Chemicals, materials, samples and instruments

The HPLC-grade methanol (MeOH, 99.8 %), acetonitrile (ACN, 99.8 %) and ethanol absolute anhydrous (EtOH, 99.9 %) were purchased from Carlo Erba (Italy). Sodium chloride (NaCl, 99.9 %), sodium hydroxide (NaOH, 98.0 %) and propan-1-ol (1-PrOH, 99.8 %) were obtained from AnalaR BDH Chemicals (UK). Propan-2-ol (2-PrOH, 99.9 %) was obtained from Fisher Scientific (UK). The hydrochloric acid (37.0 %) and sodium carbonate (99.5%) were purchased from Riedel-de Haën (Germany). Ultra-pure water was obtained from Milli-Q water purification systems (USA). The sorbent phases used for the preparation of the microextraction devices were Strata®-X (N-vinylpyrrolidone and divinylbenzene copolymer; 33 μm particle size, 800 m2 g−1 surface area, pH 1–14 stability) and Ciano (silica and ciano copolymer, particle size 55 μm, surface area 500 m2 g−1, pH 1–14 stability) from Phenomenex, USA; LiChrolut® EN (ethylvinylbenzene and divinylbenzene copolymer; particle size 40–120 μm, surface area 1200 m2 g−1, pH 1–13 stability) from Merck Millipore, Germany; HLB (N-vinylpyrrolidone and divinylbenzene copolymer, 30–60 μm particle size, 810 m2 g−1 surface area, pH 0–14 stability) from Waters, USA); and ENVI™-18 (octadecyl silica polymer, 45 μm particle size, 475 m2 g−1 surface area) from Supelco, PA, USA. Seven out of the SCs selected as model compounds used in this work were previously isolated and characterized from products supplied by the Scientific Police from Portuguese Criminal Police (Lisbon, Portugal, which also provided a standard of cumyl-5F-PINACA) [24]. All compounds were characterized by nuclear magnetic resonance (NMR). Briefly, an aliquot of each SC (10 to 15 mg) was dissolved in CDCl3 for NMR structural analysis. 13C NMR (100.6 MHz) and 1H NMR (400.1 MHz) were recorded on a Bruker Avance spectrometer. The chemical shifts were expressed as ẟ values and referenced to the residual solvent peak (CDCl3, ẟH=2.50, ẟC=39.5). Units of Hertz (Hz) were used for reporting the coupling constants. 1D (1H, 13C APT) and 2D (COSY, HMBC and HSQC) NMR experiments showed unequivocal assignments of all proton and carbon signals. All NMR results obtained were similar to previous literature reported data [24,25,26,27,28,29].

Blank assays were performed in oral fluid samples provided by the academic community (volunteers), obtained in 2018 from 10 individuals (5 female and 5 male) by passive drooling into polypropylene cryovials [6]. It was requested that the subjects did not eat, drink or smoke for at least 10 min before the samples collection. Additionally, the volunteers guaranteed that no drugs of abuse were consumed for at least 24 h prior to sampling. Assays performed on real oral fluid samples were also obtained from the academic community (volunteers) following the same procedure as above, but without the restrictions used for the blank assays. For non-disclosure purposes, the analysis was performed without any information from the donor. After collection, the vials were kept at −20 °C, to avoid drug losses, as previously demonstrated [30]. When possible, the samples were analysed on the same day. The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the Faculdade de Ciências da Universidade de Lisboa (‘Comissão de Ética para Recolha e Proteção de Dados’) nr 02/2020.

The HPLC-DAD analyses were carried using similar equipment and strategy as already described in the literature [20] with a few modifications described below. In this specific case, we employed a Kinetex C18 column, 150.0 × 4.6 mm, 2.6 µm particle size (Phenomenex, Torrance, USA). The samples were analysed using a mobile phase consisting of an isocratic mixture of water and ACN (20/80 %, v/v). The detector was set at 302 nm, and the injection volume was 20 µL.

Experimental setupPreparation of the BAµE devices

The BAµE devices were prepared within the laboratory in accordance with existing research literature [31, 32]. Each bar, measuring 7.5 mm in length and 3 mm in diameter, underwent a meticulous process of coating with powdered polymeric phase, utilizing a suitable adhesive film. Subsequently, they were securely stored at room temperature in a hermetically sealed glass flask. The microextraction devices were intentionally designed for single use, owing to their affordability, ease of preparation, and to mitigate any potential carry-over effects. Prior to utilization, the BAµE devices were subjected to a thorough cleaning regimen involving the use of MeOH and ultra-pure water.

BAµE-µLD method development

The optimization experiments were performed in glass flasks having 10 mL of ultra-pure water spiked with the target SC to get a concentration 20.0 µg L−1. Afterwards, a BAµE device coated with a specific sorbent material and a stir bar were introduced in the flask. The assays were performed at room temperature (25 ºC in a multipoint agitation plate (Variomag H+P Labortechnik Multipoint 15, Oberschleissheim, Germany). To optimize the microextraction method, several parameters affecting its performance were evaluated. This included coating selectivity, stirring rate (750, 1000 and 1250 rpm), equilibrium time (1, 2, 3, 4 and 16 h), organic modifier (MeOH; 5, 10 and 15 %, v/v), matrix pH (2.0, 5.5, 8.0 and 11.0), and ionic strength (NaCl; 5, 10 and 15 %, w/v). After the microextraction process, the BAµE devices were taken out of the flasks with clean tweezes and put into glass vial inserts containing 100 μL of the back-extraction solvent. This was followed by ultrasonic treatment (Branson 3510, Zurich, Switzerland) at room temperature. For microliquid desorption (µLD) stage, several solvents were tested to achieve the best results. These included ACN, MeOH, EtOH, 1-PrOH and 2-PrOH, at several periods of sonication time (5, 15, 30, 45 and 60 min). Afterwards, the devices were removed from the inserts with clean tweezes, and the vials were sealed and placed on the autosampler for HPLC-DAD analysis. In supplementary material S1, we provide a simplified schematic diagram of the proposed experimental procedure (Figure S1).

BAµE-µLD/HPLC-DAD validation assays

For the validation assays, oral fluid samples were subjected to homogenization by vortexing for 30 s, followed by the addition of ACN for protein precipitation [33] at a volumetric ratio of 1:2 (ACN/sample). Subsequently, the samples were vortexed again for 30 s and then centrifuged at 4000 rpm for 10 min using a Hermle Z 300 centrifuge (Germany). Next, 750 µL of the resulting supernatant was transferred to a sampling flask, and 8.5 mL of ultra-pure water, and 1 mL of MeOH was added. This was followed by the implementation of the optimized methodology using BAµE-µLD/HPLC-DAD.

Several parameters were evaluated for the purpose of validation, including sensitivity, selectivity, linearity, precision, accuracy and recovery yields, as described in the relevant literature [33,34,35]. Unless otherwise specified, all validation assays were conducted in triplicate.

Linearity, sensitivity and selectivity

The linearity was checked by spiking oral fluid samples (before ACN addition) with a standard mixture to get a final concentration in between 20.0 and 2000.0 µg L−1 (ten points), followed by using the optimized methodology. The results enabled the production of a least-squared regression plot. The acceptance criteria were that all the points in the curve should have a relative deviation ≤ 15.0 % from its nominal concentration (relative residuals, RR) and that the determination coefficient (r2) should be ≥ 0.99. The limits of detection (LODs) and quantification (LOQs) estimated with a signal-to-noise ratio (S/N) of 3/1 and 10/1, respectively, using the same approach described above. The selectivity was assessed by evaluating the interferences that could occur due to endogenous materials in the oral fluid samples. This was accomplished by applying the optimized methodology to ten non-spiked samples (ten drug-free volunteers) and by checking the resulting chromatograms for any interfering compounds, especially at the retention times of each compound.

Accuracy and precision

Method intraday (n = 6, 5 consecutive days) and interday (n = 6) accuracies and precisions were assessed by spiking oral fluid samples (before ACN addition) to get a final concentration of 50.0, 300.0 and 1000.0 µg L−1. Accuracy was calculated as bias:

$$Bias=\frac \times 100$$

where E is the experimentally determined concentration and T is the expected concentration for spiked samples. The interday and intraday precisions were calculated as relative standard deviation (RSD) from the interday and intraday accuracy assays, respectively. The acceptance criteria for accuracy and precision were that bias and RSD should be ≤ 15.0 %, respectively.

Recovery yields

The recovery assays were performed as described in the literature [34,35,36,37]. In brief, two distinct sets of samples, labelled A and B, were prepared at varying concentrations of 50.0 and 300.0 µg L−1. Set A, comprising six samples, represented a pure standard mixture. Set B, also consisting of six samples, was created by spiking oral fluid samples obtained from different drug-free volunteers with the targeted SCs prior to microextraction. Recovery was determined by evaluating the absolute peak area ratio between the two sets. Furthermore, to account for potential variations resulting from analysing different sources, the relative standard deviation (RSD) values were calculated. An RSD value of ≤ 20.0 % was deemed acceptable.