Cannabinoids assessment in plasma and urine by

Anal Bioanal Chem DOI 10.1007/s00216-016-0046-3

RESEARCH PAPER

Cannabinoids assessment in plasma and urine by high performance liquid chromatography–tandem mass spectrometry after molecularly imprinted polymer microsolid-phase extraction Juan Sánchez-González 1 & Rocío Salgueiro-Fernández 1 & Pamela Cabarcos 2 & Ana María Bermejo 2 & Pilar Bermejo-Barrera 1 & Antonio Moreda-Piñeiro 1

Received: 12 July 2016 / Revised: 15 September 2016 / Accepted: 21 October 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract A molecularly imprinted polymer (MIP) selective for cannabinoids [Δ9-tetrahydrocannabinol (Δ9-THC), 11nor-9-carboxy-Δ9-tetrahydrocannabinol (Δ9-THC-COOH), and 11-hydroxy-Δ9-tetrahydrocannabinol (Δ9-THC-OH)] has been synthesized, fully characterized, and applied to the assessment of plasma and urine analysis of marijuana abuse by high performance liquid chromatography–tandem mass spectrometry (HPLC-MS/MS). Δ9-THC-COOH was used as a template molecule, whereas ethylene glycol dimethacrylate (EGDMA) was used as a functional monomer, divinylbenzene (DVB) as a cross-linker, and 2,2′-azobisisobutyronitrile (AIBN) as an initiator. The prepared MIP was found to be highly selective for cannabinoids typically found in blood and urine, and also for cannabinol (CBN) and cannabidiol (CBD). MIP beads (50 mg) were loaded inside a cone-shaped device made of a polypropylene (PP) membrane for microsolid-phase extraction (μ-SPE) in batch mode. Optimum retention of analytes (0.1 to 1.0 mL of plasma/urine) was achieved by fixing plasma/urine pH at 6.5 and assisting the procedure by mechanical shaking (150 rpm, 40 °C, 12 min). Optimum elution conditions implied 2 mL of a 90:10 methanol/acetic acid and Electronic supplementary material The online version of this article (doi:10.1007/s00216-016-0046-3) contains supplementary material, which is available to authorized users. * Antonio Moreda-Piñeiro [email protected] 1

Department of Analytical Chemistry, Nutrition and Bromatology, Faculty of Chemistry, Universidade de Santiago de Compostela, Avenida das Ciencias s/n, 15782 Santiago de Compostela, Spain

2

Department of Pathologic Anatomy and Forensic Sciences, Faculty of Medicine, Universidade de Santiago de Compostela, Rúa de San Francisco s/n, 15782 Santiago de Compostela, Spain

ultrasound extraction (35 kHz, 325 W) for 6 min. Good precision was assessed by intra-day and inter-day assays. In addition, the method was found to be accurate after intra-day and inter-day analytical recovery assays and after analyzing control serum and urine control samples. The limits of quantification were in the range of 0.36–0.49 ng L−1 (plasma analysis) and 0.47–0.57 ng L−1 (urine analysis). These values are low enough for confirmative conclusions regarding marijuana abuse through blood and urine analysis. Keywords Molecularly imprinted polymer . Microsolid-phase extraction . Plasma . Urine . Cannabinoids . High performance liquid chromatography–tandem mass spectrometry

Introduction As stated by the European Monitoring Centre for Drug and Drug Addiction (EMCDDA) in its 2015 report, cannabis is the most commonly used illicit drug among all age groups in Europe [1]. Cannabis abuse assessment is therefore an important activity in the toxicological-forensic laboratory, mainly for demonstrating recent marijuana usage in cases of driving under the influence of drugs (DUID) [2]. It has been frequently shown that 11-nor-9-carboxy-Δ9-tetrahydrocannabinol (Δ9-THC-COOH) and 11-hydroxy-Δ9-tetrahydrocannabinol (Δ9-THC-OH) are found in blood and urine as major metabolites [3]. These two forensically relevant metabolites together with Δ9-tetrahydrocannabinol (Δ9-THC), the major psychoactive constituent of marijuana, are commonly identified/ quantified in blood (plasma/serum) when monitoring cannabis abuse [2, 4–11]. Analyses are typically performed using plasma or serum after centrifugation and/or protein precipitation

J. Sánchez-González et al.

with acetonitrile (ACN) [2, 4, 5, 7, 8, 10, 11], although some recent developments based on liquid-liquid extraction (LLE) [6] and the dried blood spot technique [9] have focused on whole blood. The complexity of whole blood (plasma/serum) and the low levels of Δ9-THC and metabolites, mainly after a post-smoking time of 24 h (heavy abusers) and after 2 h (occasional abusers) [6], lead to a pre-concentration step necessary before analyte determination. In addition to LLE procedures [6] and the DBS technique [9], solid phase extraction (SPE), involving several commercial C18-based adsorbents, is widely used for cannabinoid isolation/pre-concentration from blood [2, 4, 5, 7, 8, 10, 11]. Advances in SPE usually refer to not only the development of on-line SPE and micro-SPE (μ-SPE) procedures but also to improvements in the selectivity of adsorbents used for SPE/μSPE. Recently, on-line SPE (a polar reverse phase adsorbent) [4] and μ-SPE (small amount of C18 as an adsorbent contained into modified tips) [8] have been proposed for cannabinoid assessment in blood. Other miniaturization approaches such as μ-SPE proposed Basheer et al. [12] have consisted on holding the adsorbent material in a polypropylene (PP) membrane, which hampers the diffusion of large biomolecules in the sample, but which allows the diffusion of the analytes for further interaction with the adsorbent. Cone-shaped μ-SPE devices (instead of the typically rectangular ones) [12–21] have recently been proposed for preventing the deterioration of the heat-sealing after enclosing the adsorbent into the μ-SPE device by solvents during the loading/eluting stages [30]. Most of the reported developments are focused on analyzing environmental and food samples when using adsorbents such as multiwalled carbon nanotubes (MWCNTs) [15], commercial reverse phase materials C18 [16–19], and laboratory-prepared adsorbents such as rice husk modified to silica–Fe [20], amino and urea-grafted silica gel [21], synthetic zeolite imidazolate framework 8 (ZIF-8) [22–24], SBA-15/polyaniline para-toluenesulfonic acid nanocomposite [22], and metal-organic framework (MOF) MIL101(Cr) [23]. Regarding selectivity, the use of molecularly imprinted polymers (MIPs) as adsorbents for SPE (molecularly imprinted solid phase extraction, MISPE) [24–26] has been shown to offer excellent selectivity for those compounds used as template molecules during the MIP synthesis. The high selectivity offered by MIPs is attributed to recognition cavities complementary to the template molecule in shape, size, and chemical functionality which are created during MIP synthesis, and which are then available after template molecule removal. MIP technology has been used recently in porous membrane-based μ-SPE procedures when assessing triazines in waters and foodstuffs (MWCNTs coated with MIPs) [27], commercial MIPs for ochratoxin A in coffee [28], and labprepared MIPs for hyperoside and isoquercitrin in rat plasma [29], and for cocaine and metabolites in human urine and

plasma [30, 31], and cannabinoids in urine and oral fluid [32]. Other recent developments involves the preparation of magnetic MIPs (cocaine and metabolites in urine and plasma) [33, 34], fluorescent quantum dots-MIPs for cocaine assessment in urine [35, 36], and magnetic dual-responsive MIP for bisphenol A [37]. The aim of the current work has been the synthesis of an MIP selective for Δ9-THC-COOH (template molecule used for MIP synthesis), and also for Δ9-THC-OH (another major metabolite in blood) and Δ9-THC (the major psychoactive constituent in marijuana, present in blood at very low concentrations). To the best of our knowledge, MIPs selective for recognition of cannabinoids in plasma have not yet been reported. The prepared material was enclosed in a conical shape MIP-μ-SPE device made of PP for performing target preconcentration from plasma before HPLC-MS/MS. The loading stage was assisted by mechanical stirring (orbitalhorizontal shaking) instead of conventional magnetically stirring, allowing up to 20 MIP-μ-SPE devices to be operated simultaneously. The excellent pre-concentration factor achieved, and the use of HPLC-MS/MS lead to limits of detection lower than the cutoff values in blood and urine for confirmation analysis of marijuana abuse.

Materials and methods Instrumentation The chromatographic system consisted of a Flexar FX-15 UHPLC binary chromatographic pump (Perkin Elmer, Waltham, MA, USA), a Flexar UHPLC autosampler (Perkin Elmer), and a 3200 Q TRAP LC/MS/MS system (ABSciex, Concord, Canada). Chromatographic separations were performed with a Zorbax Eclipse Plus C18 (100 mm length × 4.6 mm i.d., 3.5 μm particle diameter) reverse phase column (Agilent Technologies, Santa Clara, CA, USA), coupled to a Phenomenex C8 guard column (4 mm length × 3.0 mm i.d) from Phenomenex (Torrance, CA, USA). Temperature control of the column was performed with a GECKO 2000 column heater (temperature control from 30 to 80 °C) from Amchro GmbH (Hattersheim, Germany). Batch MIP-μ-SPE procedure (loading stage) was performed in a Rotabit orbital-rocking platform shaker (Selecta, Barcelona, Spain) placed inside a Boxcult temperature-controlled incubation chamber (Stuart Scientific, Surrey, UK). Elution was assisted by ultrasounds using a Raypa Model UCI-150 ultrasonic cleaner bath (35 kHz, 325 W) from R. Espinar S.L. (Barcelona, Spain). MIP synthesis was performed using a low-profile roller from Stovall (Greensboro, NC, USA) placed inside the Boxcult temperature-controlled incubation chamber. The coneshaped PP envelope containing MIP beads was heat-sealed with a TN1010 heat-sealer from Siemens (Munich,

Cannabinoids assessment in plasma and urine

Germany). MIP characterization was performed with a field emission scanning electron microscope Ultra Plus (Zeiss Oberkochem, Germany), with a Spectrum Two FT-IRUATR spectrometer from Perkin Elmer, and a Fisons CHNS/O analyser EA 1108 (Carlo Erba, Milan, Italy). Other laboratory devices were a Centromix centrifuge (Selecta), a Basic20 pH meter with a glass–calomel electrode (Crison, Barcelona, Spain), a Reax 2000 mechanical stirrer (Heidolph, Kelheim, Germany), a vacuum pump (Millipore Co., Bedford, MA, USA), an oven model 207 (Selecta), a VLM EC1 metal block thermostat and N2 sample concentrator from VLM (Leopoldshöhe-Greste, Germany), and an R-210 rotavapor equipped with a B-491 heating bath and a V-740 vacuum pump (Büchi Laboryechnik AG, Flawil, Switzerland). Reagents Ultrapure water (18 MΩ cm−1 of resistivity) was obtained from a Milli-Q purification device (Millipore Co.) Drug stock standard solutions were prepared from Δ9-THC, Δ9-THCCOOH, Δ9-THC-OH, and CBN (1000 mg L−1 each, solutions in methanol) from Celliriant (Round Rock, TX, USA). CBD (solution of 2000 mg L−1) was prepared by dissolving 10 mg of CBD (National Measurement Institute Australian Government, Sidney, Australia) in 5 mL of methanol. Deuterated Δ9-THC (Δ9-THC-d3), also 1000 mg L−1 in methanol, was from Celliriant. Δ9-THC-COOH powder, used as a template when synthesizing MIP, was from Lipomed (Arlesheim, Switzerland). Other drug standard solutions used for cross-reactivity studies were cocaine (COC, 1000 mg L−1 dissolved in acetonitrile), benzoylecgonine (BZE, 1000 mg L−1 dissolved in methanol), cocaethylene (CE, 1000 mg L−1 dissolved in acetonitrile), ecgonine methyl ester (EME, 1000 mg L−1 dissolved in acetonitrile), codeine (COD, 1000 mg L−1 dissolved in methanol), morphine (MOR, 1000 mg L−1 dissolved in methanol), and 6monoacetylmorphine (6-MAM, 1000 mg L−1 dissolved in acetonitrile), also from Cerilliant. Ethylene glycol dimethacrylate (EGDMA), used as a monomer, and 2,2′azobisisobutyronitrile (AIBN), used for free-radical polymerization, were purchased from Fluka (Buchs, Switzerland). Divinylbenzene (DVB), used as a cross-linker, was from Sigma-Aldrich. The ACCUREL® PP membrane was from Membrana (Wuppertal, Germany). Acetonitrile and methanol (supragradient HPLC grade), formic acid (98 %), neutral alumina, and sodium hydroxide were from Merck (Darmstadt, Germany). Potassium dihydrogen phosphate was from BDH (Poole, UK). Toluene, 2-propanol, ammonium hydroxide, and acetic acid 96 % (m/m) were from Panreac (Barcelona, Spain). BTMF 1/11-B (drugs of abuse in serum, lyophilized) control material and FDT +25 % (drugs of abuse, drugs and alcohol in urine, lyophilized) control material (ACQ Science,

Rottenburg-Hailfingen, Germany) were used to assess accuracy. Other consumables used were: Durapore 0.20 μm membrane filters (Millipore), cellulose extraction thimbles (Filtros Anoia, Barcelona, Spain), and 0.20 μm cellulose acetate syringe filters (LLG, Meckenheim, Germany). Plasma and urine samples Plasma and urine samples from occasional and chronic cannabis users were obtained from an addiction centre in Santiago de Compostela (Spain). Drug-free plasma samples (used for method validation) were obtained from the Blood Bank of Santiago de Compostela, whereas drug-free urine samples used for method validation were obtained from laboratory staff volunteers. Plasma was obtained from whole blood by centrifugation (4000 rpm, 20 min) and kept at 4 °C. Urine samples were kept at −20 °C when necessary. Synthesis of MIP particles The pre-polymerization mixture consisted of 0.123 g of solid Δ9-THC-COOH (template) and 67.0 μL of EGDMA (monomer) dissolved in 25 mL of a 3:1 acetonitrile/toluene mixture into 30 mL glass test tubes. The template-monomer stoichiometry is therefore 1:1. The template-monomer was allowed to s elf-assemble overnight in the dark. Polymerization was then achieved by adding DVB (1.27 mL) and AIBN (0.081 g) to the pre-polymerization mixture, and after stirring (1 min) and purging with argon (5 min), sealed tubes were placed in the low-profile roller (33 rpm on its long axis) inside a temperature-controllable incubator chamber. The temperature was first ramped from room temperature to 60 °C during 2 h, and then maintained at 60 °C for 24 h. A scheme regarding the preparation process is shown in Fig. 1A. The synthesized material was vacuum filtered, washed with acetonitrile (three rinsing stages with 20 mL of acetonitrile each one), and oven-dried (40 °C) overnight. Non-imprinted polymers (NIPs) were also prepared in the same way as MIPs but without adding the template (Δ9-THCCOOH). Δ9-THC-COOH (template) was removed from the synthesized MIP by Soxhlet extraction (two 24-h cycles with 200 mL of methanol/water/acetic acid, 85:10:5). Finally, MIP beads were rinsed with ultrapure water and oven-dried at 40 °C for 12 h before use. Preparation of the MIP-μ-SPE device The MIP-μ-SPE device consisted of a narrow cone of approximately 4.0 cm in height with only one edge (upper part of the device) heat-sealed as described previously (Fig. 1B) [30]. Approximately 50 mg of MIP was enclosed in the bottom part of the cone. The advantage of this configuration is the avoidance

J. Sánchez-González et al. Fig. 1 Schematic representation of MIP synthesis (A) and MIP-μSPE procedure (B)

EGDMA

(A)

O O

O O

9-THC-COOH

DVB / AIBN

O O O

(B)

O

EGDMA

N2 evaporation to dryness

Sample (plasma/urine)

of heat-sealing degradation by methanol (used in the eluting mixture). Prepared MIP-μ-SPE devices were previously conditioned by sonication with 5 mL of 0.1 M/0.1 M KH2PO4/NaOH buffer solution (pH 6.0) for 10 min, and stored soaked in this buffer solution. MIP-μ-SPE procedure Plasma and urine samples from cannabis abusers (0.1 to 1.0 mL) were placed into 25 mL flasks and were diluted to 5.0 mL with a 0.1 M/0.1 M KH2PO4/NaOH buffer solution (pH 6.0). The mixtures were fortified with the internal standard solution (Δ9-THC-d3, 35 μg L−1), and the pH was tested/adjusted at 6.0 (addition of a small volume of the 0.1 M/0.1 M KH2PO4/NaOH buffer solution, pH 6.0). The conditioned MIP-μ-SPE device was placed into the buffered sample, and the flasks were transferred to the shaker inside the incubator chamber (40 °C) and mechanically stirred (orbital-horizontal shaking) at 150 rpm for 12 min (the shaker accepts up to 20 flasks). After analyte retention, the MIP-μ-SPE device was removed with tweezers and placed into a clean 25 mL flask containing 5 mL of 0.1 M/0.1 M KH2PO4/NaOH buffer solution at pH 6.0 for rinsing (ultrasound assistance, 37 kHz, 325 W, 8 min). After discharging the rinsing wastes, 2 mL of methanol/aqueous acetic acid 90:10 was added for elution (sonication, 37 kHz, 325 W) for 6 min. The described MIP-μ-SPE procedure was the same when performing method validation, although 5.0 mL of drugfree plasma/urine samples containing the internal standard (Δ9-THC-d 3 at 35 μg L − 1 ) was spiked at several

Loading

HPLC-MS/MS

Elution

concentration levels (see section BLiquid chromatography–tandem mass spectrometry measurement^). For all cases, the eluates were evaporated to dryness (stream of N2, 40 °C) and re-dissolved with 100 μL of mobile phase [90 % of 0.1 % (v/v) formic acid in acetonitrile and 10 % of 0.1 % (v/ v) formic acid i n w ater]. The preconcentration factor achieved was 50 by assuming a maximum volume of 5 mL of plasma/urine sample. After elution, the MIP-μ-SPE devices were soaked in 0.1 M/0.1 M KH2PO4/NaOH (pH 6.0) buffer solution for reuse. Preliminary experiments (performed at the beginning of the absorption/elution conditions optimization process), consisting of carrying out a second ultrasound elution and analyzing the extract, led to negligible analyte signals. These findings prove an efficient analyte elution and the possibility of reuse the MIP-μ-SPE devices (at least 20 retention/eluting cycles can be performed without losing the retention efficiency and without PP membrane damage). Liquid chromatography–tandem mass spectrometry measurement Separations were performed under isocratic conditions using a mobile phase consisting of 90 % of 0.1 % (v/v) formic acid in acetonitrile and 10 % of 0.1 % (v/v) formic acid in water. The flow rate was set at 0.6 mL min−1 during the first 4 min and was then increased up to 0.7 mL min−1 during the following 5 min (the chromatographic time was 9 min). Retention times were 2.64 min for Δ9-THC-COOH, 2.74 min for Δ9-THC-OH, and


4.91 min for Δ9-THC and Δ9-THC-d3. Data acquisition parameters (multireaction monitoring mode, MRM) and optimized ion source potentials and collision energies for each MRM transition (positive electrospray ionization) are listed in Table S1 (see Electronic Supplementary Material, ESM). At least two precursor ion→product ion transitions were monitored for each analyte (ESM Table S1) for guaranteeing the specificity of measurements. However, the precursor ion→product ion transitions exhibiting the highest sensitivity were finally used for determinations. The standard addition technique was used for performing the measurements. The standard addition curves were prepared by spiking in duplicate 5.0 mL of drug-free plasma/ urine aliquots with the instrumental low limit of quantification (LLOQ) of the method, 5.5, 10, 12.5 μg L−1 of Δ9THC, Δ9-THC-OH, and Δ9-THC-COOH, respectively; and also with 10, 20, 35, and 50 μg L−1 of Δ9-THC, Δ9-THC-OH, and Δ9-THC-COOH, and 20, 35, and 50 μg L−1 of Δ9-THC-OH and Δ9-THC-COOH. For all cases, Δ9-THC-d3 was added as an internal standard (35 μg L−1). Taking into account the pre-concentration factor of 50, analyte concentrations varied within the 0– 2500 μg L−1 for Δ9-THC, Δ9-THC-OH, and Δ9-THCCOOH, whereas Δ9-THC-d3 was fixed at 1750 μg L−1 in the reconstituted extract. As an example, Fig. 2 shows

MRMs obtained for a positive plasma and a positive urine sample.

Results and discussion MIP characterization Characterization of the prepared material (MIP and NIP) by SEM (ESM Fig. S1) showed that both MIP and NIP consisted of agglomerates of spherical particles of approximately 2.0 μm in diameter. Regarding elemental composition, the C, H, N, and O percentages of MIP (before and after template removal) and NIP were quite similar. C percentages varied from 88.3 % (NIP) to 88.5 % (MIP before template removal), whereas O percentages were from 2.65 % (NIP) and 2.86 % (MIP before template removal). H percentages were 8.73, 8.52, and 8.48 % for MIP before template removal, MIP after template removal, and NIP, respectively. Finally, FT-IR spectra were quite similar for MIP (after template removal) and NIP (ESM Fig. S2). This is because there are no specific functional groups in the template molecule (Δ9THC-COOH), and bands attributed to Δ9-THC-COOH overlap with those exhibited by the monomer and cross-linker (present in both MIP and NIP). As shown in ESM Fig. S2, some

Urine

Plasma

9-THC tR=4.86

9-THC tR=4.95

9-THC-d3 tR=4.87 9-THC-d3 tR=4.95

9-THC-OH tR=2.68 9-THC-OH tR=2.74

9-THC-COOH tR=2.64

Fig. 2 Extracted MRM chromatograms for a positive plasma and urine sample

9-THC-COOH tR=2.57


characteristic bands were observed: 2900 cm−1 (C-H stretch), 1460 cm−1 (C-H bending), 1727 cm−1 (C=O stretch), three weak bands within the 1600–1400 cm−1 range (C=C stretch), and 1097 and 1157 cm−1 (C-O stretch). Optimization of MIP-μ-SPE for cannabinoids Preliminary experiments Analyte retention was first investigated using aqueous cannabinoid standards (10 μg L−1 each one) by applying loading and eluting conditions previously established for cocaine and metabolite extraction from urine [30] by using a selective MIP for cocaine. Low recoveries were achieved, and the eluting solution first tested (75:20:5 dichloromethane/methanol/ammonium hydroxide) was replaced by a 95:5 methanol/acetic acid mixture, which led to higher cannabinoid concentrations in the recovered extracts. In addition, and as previously reported for cocaine and metabolite isolation from urine by MIP-μSPE [30], the analyte elution was favoured by assisting the elution stage by sonication (37 kHz, 325 W). Efficient analyte desorption from MIP particles under ultrasounds is attributed to the cavitation phenomenon (high local temperatures and pressure inside the liquid), and to the presence of high-speed microjets when the cavitation process occurs by asymmetric collapses (presence of solid particles suspended in a liquid) [38]. Quantitative elution by ultrasound assistance was proved by analyzing extracts after a second successive elution stage, and verifying the absence of analytes (negligible chromatographic signals at the retention times of analytes). Possibilities of developing a sample pre-treatment method for whole blood or plasma were therefore studied under nonoptimized loading conditions (pH 5.5, orbital-horizontal shaking speed of 100 rpm, room temperature, 15 min) and eluting conditions (95:5 methanol/acetic acid, sonication for 5 min), using drug-free whole blood and plasma (5 mL each) spiked with 10 μg L−1 of Δ9-THC, Δ9-THC-OH, and Δ9-THCCOOH (and also Δ9-THC-d 3 as an internal standard). Experiments in quintuplicate offered good repeatability when assessing Δ9-THC, Δ9-THC-OH, and Δ9-THC-COOH in plasma (RSD values of 5, 6, and 6, respectively), whereas repeatability diminished for whole blood (RSD values higher than 10 % for all cases). In addition, higher analytical recoveries were obtained for plasma samples (within the 25–50 % range) when compared to those obtained when using whole blood (lower than 25 %). Therefore, plasma instead of whole blood was selected for further investigations. Loading conditions Variables such as pH, orbital-horizontal shaking speed, temperature and loading time were successively investigated using drug-free plasma samples/urine (5.0 mL) spiked with

Δ9-THC, Δ9-THC-OH, Δ9-THC-COOH, and Δ9-THC-d3 (10 μg L−1). Although low sample volumes (0.1–1.0 mL) are adequate for positive samples, a volume of 5 mL was selected for guaranteeing that the optimized conditions can be applied to samples exhibiting low cannabinoids concentration (the pre-concentration factor is improved by increasing the volume of sample). All experiments were performed in duplicate, and concentrations (analytical recoveries) were assessed by using calibrations matched with 0.1 % (v/v) formic acid in ultrapure water/0.1 % (v/v) formic acid in acetonitrile (10:90). Analyte standard concentrations were up 500 μg L−1, and Δ9-THC-d3 (100 μg L−1) was used as an internal standard. To investigate the effect of the pH on the extraction efficiency, the pH of drug-free plasma/urine samples (5.0 mL) spiked with analytes and internal standards was adjusted at values within the 5.5–8.0 range by adding small volumes of 0.1 M/0.1 M KH2PO4/NaOH buffer solutions (pH at the desired value). The effect of pH on analyte retention was evaluated by fixing other loading variables (room temperature, orbital-horizontal shaking speed at 100 rpm, 15 min) and eluting conditions (95:5 methanol/acetic acid, sonication for 5 min). Results (mean analytical recoveries for two replicates) at each tested pH are shown in Fig. 3A for experiments with plasma, and in Fig. 4A for urine. A similar behaviour was observed for both sample types, and the highest recoveries for all analytes were obtained when buffering the plasma/urine samples at pHs 6.0 and 6.5 (analytical recoveries from 60 to 80 %), and the latter pH was therefore selected. The effect of mechanical shaking (orbital-horizontal shaking) was investigated by fixing the shaking speeds at 100, 120, 150, 180, and 200 rpm (experiments were performed in duplicate with spiked drug-free plasma/urine samples adjusted at pH 6.5, room temperature and 15 min for loading, and sonication for 5 min with 95:5 methanol/acetic acid for elution). Figures 3B (plasma) and 4B (urine) show that analytical recovery gradually rises when increasing the shaking speed up to 150 rpm. Speeds higher than 150 rpm have led, however, to lower analytical recoveries. Low efficiency at higher stirring speeds can be attributed to backdiffusion phenomena, as previously explained by some authors when using other microextraction procedures, such as hollow-fiber microextraction [39], electro-mediated microextraction [40], and magnetic MIP-μ-SPE [33]. Therefore, an orbital-horizontal shaking speed of 150 rpm was selected since higher yields are obtained without backdiffusion phenomena. Regarding loading temperature (Figs. 3C and 4C, for plasma and urine, respectively), similar extraction efficiencies were obtained at low temperatures (20, 30, and 40 °C), although a slightly higher Δ9-THC recovery was obtained when loading at 40 °C. Therefore, loading temperature was finally set at 40 °C.


(A)

9-THC 9-THC-OH 9-THC-COOH

80 60 40 20 0

5.5

6.0

6.5

7.0

7.5

100

Analytical recovery (%)

100


Fig. 3 Effect of pH (A), orbitalhorizontal shaking speed (B), loading temperature (C), and loading (extraction) time (D) on the analytical recovery of Δ9THC, Δ9-THC-OH, and Δ9THC-COOH from plasma samples

(B)

60 40 20 0

8.0

100

pH

(C)


80 60 40 20 0

20

30

40

120

50

100

(D)

40 20 0

60

3

6

100

60 40 20 6.5

7.0

7.5

40 20


60 40 20 30

40

50

Loading temperature (°C)

60

100


(C)

20

100

120

150

180

200

Orbital-horizontal shaking speed (rpm)

80

0


60

0

8.0

(B)

80

pH 100

15

As previously mentioned in section BPreliminary experiments,^ analyte elution from MIP particles was favoured by using methanol/acetic acid mixtures rather than the proposed eluting solution for cocaine (75:20:5


6.0

12

Elution conditions

(A)

5.5

10

Loading time (min)

cavities implies fast sorption kinetics. This time is quite lower than that required when using other MIP-based SPE methods such as MISPE pills (cannabinoids isolation from urine and oral fluid) which imply an incubation period of 72 h before analytes desorption [32].

80

0

200

60



100


Fig. 4 Effect of pH (A), orbitalhorizontal shaking speed (B), loading temperature (C), and loading (extraction) time (D) on the analytical recovery of Δ9THC, Δ9-THC-OH, and Δ9THC-COOH from urine samples

180


80

Loading temperature (°C)

The effect of the loading time on the analyte recovery was investigated by performing the loading stage for times within the 3–15-min range. As shown in Figs. 3D and 4D, extraction efficiency is gradually improved when increasing the loading time from 3 to 12 min. However, loading times higher than 12 min worsen the extraction efficiency. Bad recoveries at high loading times can also be explained by back-diffusion phenomena. These findings are similar to those observed when using high shaking speeds, and large shaking speeds and loading times can lead to analyte desorption. Therefore, the loading time was fixed at 12 min. The short time for analyte interaction with MIP recognition

150

Orbital-horizontal shaking speed (rpm) Analytical recovery (%)


100


80

(D)


80 60 40 20 0

3

6

10

12

Loading time (min)

15


dichloromethane/methanol/ammonium hydroxide) when also using MIP-μ-SPE sample pre-treatments [30]. Therefore, the methanol/acetic acid ratio was investigated in several experiments performed by fixing the selected loading conditions (section BLoading conditions^) and by using 2 mL of the eluting solution tested under sonication (37 kHz, 325 W) for 5 min. As shown in Fig. 5A (plasma) and 5B (urine), low analyte recoveries were obtained when using methanol (methanol/acetic acid ratio of 100:0), whereas the elution efficiency was increased when adding acetic acid in the elution mixture. The highest recoveries were obtained when the acetic acid percentage was within the 10–20 % range (an acetic acid percentage of 10 % was finally proposed). The effect of the eluting time for analyte release from MIP particles was tested (Fig. 5C, D), and the highest analytical recoveries were obtained for eluting times higher than 5 min for plasma experiments, and higher than 6 min for urine (sonication for times lower than 5–6 min led to low recoveries for all analytes). Therefore, an elution time of 6 min was finally proposed for speeding up the whole process.

High analytical recoveries (extraction efficiency) were obtained for Δ9-THC-COOH (template molecule) and also for other cannabinoids such as Δ9-THC, Δ9-THC-OH, CBD, and CBN (Table 1) when using MIP-μ-SPE, whereas analytical recoveries lower than 25 %, or even negligible, were obtained for NIP-μSPE (Table 1). These findings prove that analyte interaction with MIP particles inside the PP membrane are specific (the interaction occurs through the imprinted cavity in the MIP particles). In addition, the prepared MIP can be used for recognizing the template (Δ9-THC-COOH) as well as other cannabinoids (Δ9THC, Δ9-THC-OH, CBD, and CBN) since higher distribution ratios and lower selectivity factors were calculated (Table 1). Regarding other drugs and metabolites such as COC, BZE, EME and CE (cocaine abuse), and MOR, COD, and 6-MAM (heroin abuse), low analytical recoveries and distribution ratios (and hence high selectivity coefficients) were obtained when using MIP-μ-SPE and NIP-μ-SPE. This fact implies that these substances are not retained by the MIP particles by neither specific interactions nor non-specific interactions (adsorption). In addition, absorption onto NIP particles is negligible, mainly for MOR, COD, and 6-MAM.

Imprinting effect and cross-reactivity studies Sorption capacity of MIP and reusability The imprinting effect and cross-reactivity studies were performed by using MIP-μ-SPE and NIP-μ-SPE devices, and drug-free plasma samples (5.0 mL) spiked with analytes (Δ9-THC, Δ9THC-OH, and Δ9-THC-COOH) and other drugs/metabolites as listed in Table 1, all at a concentration of 25 μg L−1. Parameters such as the extraction efficiency (analytical recovery), the distribution ratio (D), and selectivity coefficient (SΔ9-THC-COOH/D), defined as shown in Table 1, were calculated for a set of experiments in triplicate using both MIP-μ-SPE and NIP-μ-SPE devices under optimized conditions (section BLoading conditions^ and BElution conditions^).

(A)

80 60 40 20 0

60/40

100

70/30

80/20

90/10


60 40 20 3

5

6

7

Eluting time (min)


80 60 40 20 60/40

70/30

80/20

90/10

100

Methanol/Acetic acid ratio

80

0

(B)

100

0

100

Methanol/Acetic acid ratio

(C)


120


8

120



100


Fig. 5 Effect of methanol/acetic acid ratio on the analytical recovery of Δ9-THC, Δ9-THCOH, and Δ9-THC-COOH from plasma (A) and urine (B) samples; and effect of eluting time on the analytical recovery of Δ9-THC, Δ9-THC-OH, and Δ9-THCCOOH from plasma (C) and urine (D) samples

Sorption capacity was evaluated by treating two MIP-μ-SPE devices (50 mg of MIP inside) with several Δ9-THC-COOH standard solutions (5.0 mL) of increasing concentrations (from 50 to 1000 μg L−1 in increments of 50 μg L−1) under optimized loading/eluting conditions. Quantitative recoveries (analytical recoveries within the 90–110 % range) were observed, except when loading the more concentrated solutions (concentrations higher than 550 μg L−1). The retention capacity of the polymer (defined as the maximum amount of template retained from 1 g of material) was therefore calculated to be 0.050 mg g–1.


(D)

100 80 60 40 20 0

3

5

6

7

Eluting time (min)

8

Cannabinoids assessment in plasma and urine Table 1 Extraction efficiency (%), distribution ratios (D), and selectivity coefficients (SΔ9-THC-COOH/D) for MIP-μ-SPE and NIP-μSPE: Δ9-THC-COOH (11-nor-9-carboxy-Δ9-tetrahydrocannabinol), Δ9-THC-OH (11-hydroxy-Δ9-tetrahydrocannabinol), Δ9-THC (Δ9tetrahydrocannabinol), CBD (cannabidiol), CBN (cannabinol), COC (cocaine), BZE (benzoylecgonine), EME (ecgonine methyl ester), CE (cocaethylene), 6-MAM (6-monoacetylmorphine), COD (codeine), MOR (morphine) Extraction efficiency (%)a

Distribution ratio (D)b

Selectivity coefficient (SΔ9-THC-COOH/D)c

Δ9-THC-COOH Δ9-THC-OH

94 90

15 9.1

– 1.8

Δ9-THC CBD

92 87

14 7.0

1.4 2.2

CBN

92

11

1.4

COC

23

0.29

53

BZE EME

0 16

0 0.19

∞ 82

CE 6-MAM COD

9 1 2

0.097 0.014 0.019

169 826 776

MOR NIP-μ-SPE

2

0.019

847

Δ9-THC-COOH Δ9-THC-OH Δ9-THC

0 25 26

– 0.34 0.35

– 46 44

CBD CBN

20 22

0.26 0.30

61 69

COC BZE EME

15 0 4

0.18 0 0.039

85 ∞ 401

CE 6-MAM COD

17 2 2

0.23 0.019 0.020

117 1088 833

MOR

2

0.018

834

MIP-μ-SPE

A1 = amount of analyte in aqueous solution at equilibrium. A2 = amount of analyte enriched by MIP/NIP at equilibrium. AT = total amount of analyte used in extraction. DΔ9-THC-COOH = distribution ratio for Δ9-THC-COOH (template). DD = distribution ratio for D (D = Δ9-THC-OH, Δ9-THC, CBD, CBN, COC, BZE, EME. CE, 6-MAM, COD, MOR) a

% = (A2 / AT) × 100

b

D = (A2 / A1)

c

SΔ9-THC-COOH/D = DΔ9-THC-COOH / DD

Regarding the reusability of the MIP-μ-SPE device, three independent MIP-μ-SPE devices were successively used for pre-concentrating the analytes from a drug-free plasma sample spiked at three concentration levels (10, 35, and 50 μg L−1) during the performance of the inter-day/intra-day precision/ analytical recovery assays (an MIP-μ-SPE for each concentration level when preparing the standard addition graph). Each

MIP-μ-SPE device was therefore re-used 19 times when assessing each spiked concentration level. As shown in Fig. S3 (ESM), quantitative analytical recoveries were obtained after 19 successive loading/elution cycles for all analytes and MIP-μ-SPE devices, which proves that each MIP-μ-SPE device can be re-used at least 19 times. Calibration. Evaluation of matrix effect Six different calibration curves covering analyte concentrations of 50, 100, 250, and 500 μg L−1, and using internal standard (Δ9-THC-d3 at 250 μg L−1) were prepared in several days when optimizing MIP-μ-SPE conditions. Slopes of the calibration curves, expressed as mean ± standard deviation, are listed in Table 2. In addition, 11 standard addition curves were prepared with drug-free plasma and drug-free urine when studying interday and intra-day assays. These standard addition calibration consisted of spiking drug-free plasma/urine samples (5.0 mL) with the deuterated analogue (Δ9-THC-d3 at 35 μg L−1), and analyte (Δ9-THC, Δ9-THC-OH, and Δ9-THC-COOH) standards (10, 20, 35, and 50 μg L−1 each), followed by MIP-μSPE. Standard addition graphs, as well as other analytical performances (see next sections), have been established by spiking 5.0 mL of drug-free plasma/urine samples. Validation of the method is therefore performed with a sample volume high enough for guaranteeing the applicability of the method to samples exhibiting low cannabinoid concentrations. The slopes (also expressed as mean ± standard deviation) for standard addition graphs from plasma and urine are also listed in Table 2. It can be concluded that there is good repeatability for both the calibration and standard addition calibration curves. In addition, good linearity of the method (both calibration and standard addition curves) was also obtained because the regression coefficients were higher than 0.995 in all cases. However, matrix effect is important due to the high differences in the average slopes of both calibration methods (average slopes for calibration and standard additions were found to be statistically different after applying the Multiple Range Test at a 95 % confidence interval). Therefore, the standard addition calibration must be used for avoiding matrix effect and obtaining accurate results. Limit of detection and limit of quantification The limit of detection (LOD), the low limit of quantification (LLOQ), and the limit of quantification (LOQ) were established using the following equations 3σ m 5σ LLOQ ¼ m 10σ LLOQ ¼ m

LOD ¼

J. Sánchez-González et al. Table 2 Mean slopes of calibration and standard addition with plasma and urine, and LOD, LLOQ, and LOQ values

Mean ± Sd (calibration)a Mean ± Sd (standard addition plasma)b Mean ± Sd (standard addition urine)b Plasma LOD (ng mL–1)c

Δ9-THC

Δ9-THC-OH

Δ9-THC-COOH

0.521 ± 0.0434 1.00 ± 0.136 1.50 ± 0.103

0.328 ± 0.0738 0.851 ± 0.108 0.657 ± 0.075

0.362 ± 0.0352 0.851 ± 0.178 0.509 ± 0.080

0.11

0.12

0.15

LLOQ (ng mL–1)c LOQ (ng mL–1)c

0.18 0.36

0.21 0.41

0.25 0.49

Urine LOD (ng mL–1)c LLOQ (ng mL–1)c LOQ (ng mL–1)c

0.16 0.27 0.53

0.17 0.28 0.57

0.14 0.23 0.47

a

n=6

b

n = 11

c

Pre-concentration factor of 50

were σ is the standard deviation of 11 measurements of a blank (a drug-free plasma or a drug-free urine sample), and m is the mean slope of the standard addition graphs when using drug-free plasma and drug-free urine sample. Table 2 lists LOD/LLOQ/LOQ values for experiments

with plasma and urine taking into account the preconcentration factor of the method (50). These values are lower than the cutoff values for cannabis abuse by whole blood analysis (0.63 ng mL−1 of Δ9-THC) and urine (10 ng mL−1 of Δ9-THC/Δ9-THCCOOH) reported

(A) Drug-free urine

Drug-free plasma

9-THC tR=4.62

9-THC tR=4.58

9-THC-d3 tR=4.20

9-THC-d3 tR=4.56

9-THC-OH tR=2.59

9-THC-OH tR=2.58

9-THC-COOH tR=2.48 9-THC-COOH tR=2.48

Fig. 6 Extracted MRM chromatograms for a drug-free plasma and a drug-free urine (A) sample; and for drug-free plasma and drug-free urine (B) samples spiked at LOQ level


(B) Drug-free urine spiked at the LOQ

Drug-free plasma spiked at LOQ level

9-THC tR=4.26

9-THC tR=4.17

9-THC-d3 tR=4.20

9-THC-d3 tR=4.16

9-THC-OH tR=2.56

9-THC-OH tR=2.57

9-THC-COOH tR=2.48

9-THC-COOH tR=2.48

Fig. 6 (continued)

by Vindenes et al. for DUID [41]. In addition, the high pre-concentration factor achieved allows very low LOQs, which are lower than those reported by other authors when using MIP-based SPE methods and HPLC-MS/MS as an analytical technique [32]. Finally, MRM chromatograms from one replicate of drug-free plasma and drug-free urine (Fig. 6A) show that any signal from potential interferences was detected for the selected precursor ion→product ion transitions. In addition, MRM chromatograms obtained from replicates of drug-free plasma and drug-free urine spiked at the LOQ levels (Fig. 6B) show intense signals. Selectivity Since a selective MIP is used as an adsorbent for μ-SPE, and tandem MRM is used for detection of the analytes, selectivity of the proposed procedure is high. Experiments based on using 11 drug-free plasma/urine samples (blanks) subjected to the MIP-μ-SPE procedure (six MIP-μ-SPE devices) did not show chromatographic signals at the selected precursor ion→product ion transitions (see Fig. 6). These findings also prove that the use of high purity chemicals in sample preparation and for performing the chromatographic separation was not found to contribute to positive blank values. In addition, results when assessing sorption capacity (section BSorption capacity of MIP and reusability^) show

the absence of carryover effects since quantitative analytical recoveries were obtained even after applying the method to 1000 μg L−1 Δ9-THC-COOH standards.

Table 3 Intra-day precision and inter-day precision (RSD/%), and intra-day analytical recovery and inter-day analytical recovery (AR/%) of the method (plasma samples) Added concentration/μg L–1 Δ9-THC 4.5 10 35 50 Δ9-THC-OH 4.5 10 35 50 Δ9-THC-COOH 4.5 10 35 50 a

Intra-day assay (n = 5)

b

Inter-day assay (n = 5)

RSD/%a

RSD/%b

AR/%a

AR/%b

8 3 5 3

11 4 5 1

95 ± 8 94 ± 3 99 ± 5 98 ± 3

104 ± 10 96 ± 4 100 ± 5 100 ± 1

8 7 6 1

10 5 4 5

97 ± 8 99 ± 7 93 ± 6 102 ± 1

97 ± 10 96 ± 5 98 ± 4 100 ± 5

8 6

11 5

103 ± 7 97 ± 6

96 ± 10 99 ± 5

6 3

4 4

93 ± 6 99 ± 3

102 ± 4 100 ± 4

J. Sánchez-González et al. Table 4 Intra-day precision and inter-day precision (RSD/%), and intra-day analytical recovery and inter-day analytical recovery (AR/%) of the method (urine samples) Added concentration/μg L–1

RSD/%a

RSD/%b

AR/%a

AR/%b

5 5

6 5

90 ± 6 90 ± 5

92 ± 6 100 ± 5

6

1

92 ± 5

100 ± 1

10

5

4

95 ± 5

93 ± 4

35 50

3 3

4 3

96 ± 3 97 ± 3

98 ± 4 101 ± 4

10 35

4 3

5 4

96 ± 4 97 ± 3

94 ± 4 100 ± 4

50

2

2

97 ± 3

101 ± 2

Δ9-THC 10 35 50 Δ9-THC-OH

Δ9-THC-COOH

a

Intra-day assay (n = 5)

b

Inter-day assay (n = 5)

Intra-day and inter-day assays. Analysis of control samples Intra-day precision and intra-day analytical recovery assays were assessed at three concentration levels (10, 35, and 50 μg L−1) for drug-free urine samples, whereas four concentration levels (4.5, 10, 35, and 50 μg L−1) were used for drugfree plasma. Each tested concentration level was performed five times, whereas the other concentration levels were replicated twice (the intra-day assay was performed in three different days for urine, and four different days for plasma, one day for each tested concentration level). RSD values after sample pre-treatment procedure and HPLC-MS/MS measurement using the standard addition calibration technique are listed in Tables 3 and 4. Good intra-day precision was proved for all concentration levels and analytes since RSD values lower than 8 and 9 % were calculated for plasma and urine, respectively. In addition, the mean analytical recovery (intra-day accuracy),

Table 5 Δ9-THC, Δ9-THCOH, and Δ9-THC-COOH concentrations in serum and urine control samples

also listed in Tables 3 and 4, shows good accuracy because analytical recoveries were within the 93 and 102 % range for plasma, and from 90 to 99 % for urine. Similarly, inter-day precision and the inter-day accuracy were also evaluated by preparing five standard addition curves in five different days using drug-free plasma/urine samples and performing each concentration level in duplicate. Results, also listed in Tables 3 and 4, show RSD values for inter-day precision lower than 11 and 8 %, for plasma and urine, respectively; and inter-day analytical recoveries between 96 and 104 % (plasma) and between 92 and 101 % (urine). Finally, accuracy was also tested by analyzing a BTMF 1/ 11-B control serum sample and FDT +25 % control urine sample, which offer ranges of referenced concentrations for Δ9-THC, Δ9-THC-OH, and Δ9-THC-COOH. Control samples were prepared according to manufacturer’s recommendations, and the reconstituted samples were subjected in triplicate to the proposed MIP-μ-SPE and HPLC-MS/MS procedure. It can be seen in Table 5 that the assessed concentrations were within the control ranges (established by using the standard deviation according to Horwitz [42]), and also within the confidence ranges (significance level of 99 %) provided by the manufacturer for plasma control. The developed MIP-μ-SPE procedure is therefore accurate.

Applications The applicability of the method was tested by analyzing several plasma and urine samples from chronic and occasional cannabis users. The samples were subjected to the optimized MIP-μ-SPE procedure in triplicate and were analysed twice by HPLC-MS/MS. Δ9-THC as well as two main metabolites (Δ9-THC-OH and Δ9-THCCOOH) were detected/quantified in all plasma samples (five plasmas). Δ9-THC concentrations varied from 2.2 to 33.0 ng mL−1, whereas Δ9-THC-OH and Δ9-THCCOOH levels were within the 0.67–14.2 and 0.81– 8.1 ng mL−1, respectively. Regarding urine samples, only

Found concentration (ng mL−1) BTMF 1/11-B control serum sample Δ9-THC 19.26 ± 0.23 Δ9-THC-OH 9.51 ± 0.12 Δ9-THC-COOH 137.01 ± 3.22 FDT +25 % control urine sample Δ9-THC