Introduction

New psychoactive substances (NPS) consist of a worldwide concern due to continuous increase in the distribution and synthesis of new compounds. In most cases, NPS are not controlled by international drug regulations, and only regional specific regulations have been established [1]. In 2019, more than 950 compounds were reported in more than a hundred countries [2]. This dynamic market aims to circumvent laws regarding the regulation of drugs of abuse [3, 4]. Most of the information related to NPS is obtained by analyzing seized materials such as powders [5], blotters [6], plant products [7] and tablets [8]. However, there is a lack of data concerning their detection in biological samples from intoxication cases [9].

Particularly regarding NPS, several difficulties in the analytical investigation of these compounds have been reported, because many precursors and metabolites remain to be discovered. Moreover, the determination of the targeted molecules can be a challenging work, because of the low concentrations of the analytes and complexity of biological matrices. Generally, a small dose (ng) of NPS can promote intense psychoactive effects [10]. Therefore, high sensitivity is generally required for analytical methodologies.

To overcome these challenging aspects, sample pretreatment is commonly one of the most critical steps for their analyses to concentrate and purify target compounds from the biological matrices, which are compatible with the analytical instruments [11, 12]. Conventionally, liquid–liquid extraction (LLE) and solid-phase extraction (SPE) have been used as sample preparation techniques. Both techniques are efficient; however, LLE employs large amounts of toxic organic solvents, and the use of SPE cartridges may promote high cost per analysis [13]. The use of the 96-well plate system can represent a turning point in semiautomated and automated analyses because it has already been successfully employed with several microextraction approaches.

The development and application of microextraction techniques can overcome these limitations. Generally, these techniques allow reduced analysis time and low amounts of solvents and sample consumption, decreasing the analysis costs. Several methodologies based on miniaturization of classical approaches have been developed [14].

An important trend regarding microextraction approaches refers to the principles of the “green analytical chemistry” in which extraction phases based on environmentally friendly solvents (deep eutectic solvents, ionic liquids, amphiphilic solvents) and sorbents can be applied [15,16,17]. Reviews describing the application of microextraction techniques in bioanalytical [11,12,13, 18] and toxicology [19,20,21] have been reported.

This review consists of fundamentals of some microextraction techniques with focus on applications for the determination of NPS in biological samples. Some definitions and classifications of NPS, as well as a brief overview of some important microextraction techniques are highlighted. Moreover, detailed discussions related to recent studies for the determination of NPS in biological matrices are conducted.

New psychoactive substances

NPS are being used as alternatives to traditional drugs of abuse, and they are often organized based on their effects (stimulants, depressants, hallucinogens, etc.) or based on chemical properties. The United Nations Office of Drug and Crime (UNODC) has defined nine different groups including aminoindanes, phencyclidine-type substances, phenethylamines, piperazines, plant-based substances, synthetic cannabinoids (SCs), synthetic cathinones, tryptamines, and other substances [2]. Other agencies such as European Monitoring Center for Drugs and Drug Addiction (EMCDDA) define some other relevant groups of substances such as opioids, benzodiazepines, arylalkylamines, arylcyclohexylamines, piperidines and pyrrolidines [22]. According to this information, it is possible to observe the diversity of these compounds which represents a great analytical challenge for laboratories.

SCs are one of the largest and most popular NPS groups, commonly marketed as a mixture of plants by soaking or spraying [23]. This class of compounds includes classical cannabinoids, which are analogues of Δ9-tetrahydrocannabinol (Δ9-THC), designed to exhibit higher effect than the original compound found in cannabis [24]. Then, a series of non-classical cannabinoids has been developed to be used in pharmaceutical applications, including cyclohexylphenols and hybrid cannabinoids [25]. However, the most popular compounds in recreational market are aminoalkylindoles, such as JWH-018, since their synthesis is simple and cheap to be performed. This group includes naphthoylindoles, phenylacetylindoles and benzoylindoles [26]. These compounds can bind to cannabinoid receptors, CB1 and CB2, to produce their effects [23].

Synthetic cathinones are another reported group of NPS. These compounds are derived from cathinone of khat plants (Catha edulis), and they can be found as powder, pill, or capsule in products named “research chemicals”, “plant food”, “bath salts” or “glass cleaner”. The chemical structure of cathinone is similar to amphetamine, with the addition of a β-keto group [27, 28]. Further modifications include changing the aromatic ring, alkyl side chain, and amino group. Synthetic cathinones promote their effects by stimulating the central nervous system (CNS) and are capable of mediating levels of dopamine, serotonin, and noradrenaline. Therefore, their effects can be compared to cocaine, amphetamine, and ecstasy (MDMA) [29].

Moreover, another group of NPS includes aminoindanes with the chemical structure based on 2-aminoindane (2-AI) which is a cyclic analogue of amphetamine. Other compounds of this class are synthetized based on modifications of 2-AI. Aminoindanes effects are associated with the mediation of serotonin release and uptake [30].

Phenethylamines consist of a large group of psychedelic substances derived from modifications of mescaline molecule, such as 2C-X series and NBOMe series. NBOMe compounds are often used to replace LSD, and their synthesis is based on the addition of a 2-methoxybenzyl group to the nitrogen atom of the core [31, 32]. Recently, modifications of the chemical structure have been performed with the addition of N-benzylhydroxy group, resulting in NBOH compounds, which aims to be a “legal” alternative to NBOMes [33]. Another class is benzodifuranyl substances, popularly known as FLY compounds, including Bromo-DragonFLY [34]. Effects include stimulant properties and hallucinogenic characteristics similar to LSD by the influence of the 5-HT2A serotonin receptor [32].

NPS that do not fit in one of the nine groups defined by UNODC are organized as "other substances". Opioids are a major issue in some countries worldwide and the main compounds are fentanyl and its derivatives, and non-fentanyl analogs such as U-47700 and AH-7921. A great concern is related to more pronounced effects caused by these compounds when compared with common opioids. The sedative effects are achieved mainly by influencing µ-opioid receptors and they are similar to those provided by morphine and heroin [35]. Other sedative effects can be achieved with the use of benzodiazepines by modulating GABAA receptors [36]. Designer benzodiazepines are often synthetized using pharmaceutical compounds as analog. They are marketed as fake versions of these medicines or as legal alternatives to avoid the need for a medical prescription [22]. The classification of mentioned NPS is shown in Table 1.

Table 1 Classification of mentioned NPS
Full size table

Liquid-based microextraction techniques

Liquid-phase microextraction (LPME) techniques require the use of few microliters of organic solvent and they can be organized in three main configurations including single-drop microextraction (SDME), dispersive liquid-liquid microextraction (DLLME) and hollow-fiber liquid-phase microextraction (HF-LPME). Other membrane-based configurations such as electromembrane extraction (EME) and parallel artificial liquid-membrane extraction (PALME) can also be mentioned as recent developments. These techniques have been developed to circumvent the limitations associated with LLE [17].

Organic solvents such as methyl t-butyl ether (MTBE), benzene, pentane, hexane, cyclohexane, methylene chloride, chloroform, and carbon tetrachloride have also been used in LPME-based approaches. However, the volume required of these chemicals is dramatically reduced when compared to LLE. Due to some properties, including toxicity, carcinogenicity and high volatility, these organic solvents have gradually been replaced by environmentally friendly alternatives [37].

Efforts are being conducted to develop analytical methodologies employing solvents with greener chemical properties, which include reduced toxicity and environmental sustainability. Among these solvents, ionic liquids, and deep eutectic solvents have shown to be of interest [38]. The solvent choice consists of an important task, and physicochemical properties such as melting and boiling points, vapor pressure, viscosity, density, surface tension, polarity, and toxicity must be considered. Moreover, the solvents should effectively interact with the analytes to provide satisfactory extractions [37].

These techniques have been applied to determine several analytes such as environmental pollutants [39,40,41], pesticides [42], food constituents [43] and pharmaceuticals [44]. Table 2 summarizes the applications of liquid-based microextraction techniques for the determination of NPS.

Table 2 Liquid-based microextraction techniques
Full size table

Dispersive liquid-liquid microextraction

DLLME was introduced in 2006 [45] for the extraction and pre-concentration of organic compounds from water samples. In this case, a mixture of extraction and disperser solvents is rapidly added to the aqueous sample, forming a cloudy solution. The disperser solvent must be miscible in both the extraction solvent and aqueous sample. Furthermore, the extraction solvent should be hydrophobic to maintain a biphasic system [38, 46]. This cloudy solution contains droplets of the extraction solvent with high surface area which allows for effective enrichment and rapid extraction kinetics. Afterwards, the sample is centrifuged allowing for the deposition of the extraction solvent. This extract is then collected by a syringe and directly injected into the analytical system or evaporated to dryness and reconstituted in an appropriate solvent before the injection. An illustration of DLLME is shown in Fig. 1.

Fig. 1
figure 1

Schematic of dispersive liquid-liquid microextraction (DLLME) technique

Full size image

The performance of DLLME approaches can be affected by both extraction and disperser solvents as well as their volumes, which must be optimized. Moreover, other variables such as salting-out effect, extraction time and sample pH can also exhibit influence on the extraction efficiency [47]. The advantages of this technique include rapid extraction, simplicity of operation and low cost. However, the main disadvantage associated with DLLME is the use of halogenated solvents. As an alternative, low-density solvents can be used and recovered by solidification. In this case, phase separation is performed by cooling the sample to allow solvent solidification [38, 45].

DLLME has been applied for the determination of NPS in biological matrices such as whole blood [48], oral fluid [49], and hair [50]. Odoardi et al. [48] reported the analysis of whole blood for the screening of 28 SCs, 12 cathinones, four amphetamines, three piperazines, two aminoindanes, three ephedrine derivatives, two β-carbolines, ketamine and methoxetamine. In this particular case, 500 µL of methanol was added to 500 µL of whole blood for protein precipitation. After a centrifugation step, an aliquot of 500 µL of the supernatant was transferred into a tube containing 1 mL of water, 0.2 g of NaCl and 100 µL of carbonate buffer (pH 9). An aliquot of 350 µL consisting of a mixture of 1:2.5 (v/v) of chloroform (extraction solvent) and methanol (disperser solvent) was added to the sample. The sample was sonicated and centrifuged, and 50 µL of the organic phase was collected, evaporated, and reconstituted in 20 µL of methanol and 80 µL of water with 0.1% formic acid. The final solution was subjected to ultra-high performance liquid chromatography–tandem mass spectrometry (UHPLC–MS/MS) analysis. Limits of detection (LODs) ranged from 0.2 to 2 ng/mL for all analytes, and linear ranges varied from 5 to 100 ng/mL for cannabinoids and 5 to 1000 ng/mL for the stimulants. The use of DLLME provided a low matrix effect for all the analytes, with efficient recoveries and no interfering from other compounds [48].

Fernández et al. [49] reported an analytical method for the determination of twenty drugs of abuse (including 9 synthetic cathinones) in oral fluid. An aliquot of 1.4 mL of methanol (used also as disperser solvent) was added to the diluted sample of oral fluid, mixed and centrifuged for 5 min to precipitate proteins. Then, the collected supernatant was mixed with 1.5 mL of water at pH 8 containing 0.1 g of NaCl and 200 µL of chloroform (extraction solvent). 5 min of sonication was used to assist the extraction step, followed by centrifugation for 5 min and collection of the organic phase. After the evaporation and reconstitution in 50 µL of mobile phase, 5 µL of the extract was analyzed by ultra-performance liquid chromatography–tandem mass spectrometry (UPLC–MS/MS). Variables such as volumes of disperser and extraction solvents, sample pH, and sonication time were optimized. In this case, limits of quantification (LOQs) ranged from 0.25 and 5 ng/mL for synthetic cathinones. Linearity ranged from 0.5 to 500 ng/mL for ethylone, MDPV and pyrovalerone; 1–500 ng/mL for ethylcathinone, ephedrine and butylone; 2.5–500 ng/mL for methylone, mephedrone and pentedrone; and 5–500 ng/mL for ethylcathinone. Intraday and interday precision were lower than 15%, and recoveries ranged between 80 and 115% for most of the compounds at low, medium, and high concentrations [49].

In another study, Vincenti et al. [50] proposed a screening method for the determination of sixty drugs of abuse including 18 SCs, 10 cathinones, and five phenethylamines in hair samples. In this work, pressurized liquid extraction (PLE) was used combined with DLLME to achieve satisfactory clean-up of the samples. First, PLE was performed for hair decontamination and analyte extraction. An aliquot of 5 mL of the PLE extracts was added with 1.2 g of NaCl, 5 mL of water, 20 µL of 1 M NaOH and 500 µL of carbonate buffer. A mixture of 200 µL of chloroform (extraction solvent) and 500 µL of 2-propanol (disperser solvent) was then added to the solution. Homogenization was achieved by sonication for 10 min, and centrifugation for 5 min was used for the sedimentation of the organic phase, with 200 µL being collected, evaporated, and reconstituted with 100 µL of methanol/water (50:50, v/v). The analyses were performed using UHPLC–high-resolution (HR)-MS/MS and allowed for LOQs ranging between 2 and 50 pg/mg. Matrix effects were minimized due to sample preparation with values between 91 and 118%, meanwhile recovery varied from 15 to 95% for all analytes. Intraday precision results were lower than 15% and interday precision results were lower than 26% considering relative standard deviation (RSD) for three concentrations [50].

Membrane-based techniques

Electromembrane extraction (EME) was developed in 2006 by Pedersen-Bjergaard and Rasmussen as an alternative configuration of HF-LPME [51]. In this approach, a supported liquid membrane (SLM) comprised of organic solvent is also used. However, the extraction process occurs with the addition of an electric potential to facilitate the migration of the analytes from the aqueous sample (donor phase) into the acceptor solution contained in the lumen of the hollow membrane. The aim of this configuration is to increase enrichment factors with shorter extraction times [51].

Selectivity in EME depends on changes in pH of the sample and acceptor solution, type of solvent used for the supported liquid membrane, and characteristics of the electric field such as direction and magnitude [52]. Generally, non-polar basic compounds (logP > 1.7) can be effectively recovered with the use of nitro aromatic solvents such as 2-nitrophenyl octyl ether (NPOE). However, for more polar basic compounds (logP < 1), addition of di-(2-ethylhexyl) phosphate or tris(2-ethylhexyl) phosphate is recommended due to their capability of acting as an ion-pairing agents [53]. Moreover, changes in pH of the donor and acceptor solutions can widely influence the ion balance of the system, affecting the migration of ions through the membrane. To promote the migration of the analytes across the SLM, these compounds should be present in ionized forms. Therefore, the pH of the donor solution must be adjusted aiming to ensure maximum ionization of the analytes [54].

The EME has recently been applied for the determination of several drugs of abuse, including NPS, in plasma samples [55]. Polyvinylidene fluoride (PVDF) filters were placed in the bottom of acceptor wells of a 96-well plate system and used as SLM. Each SLM was comprised of 3 μL of NPOE and the acceptor solution was 100 μL of 250 mM trifluoroacetic acid with dimethyl sulfoxide (1:1, v/v). Aliquots of 50 µL of plasma and 50 µL of internal standard were added to the donor wells. Experiments were carried out maintaining sample pH at 7.4 and potential of 20 V for 15 min. A scheme of this configuration is shown in Fig. 2. The variables pH, solvent ratio, extraction time and potential were optimized, and UHPLC–MS/MS was used for the analysis with ammonium formate buffer (pH 3.1) and methanol used as mobile phase at 0.5 mL/min. The method was proved to be efficient with extraction recoveries ranging from 37 to 101%, and LODs ranging from 0.0050 to 8.2 ng/mL with acceptable matrix effects (± 20%) [55].

Fig. 2
figure 2

Schematic of the electromembrane extraction (EME) technique

Full size image

The PALME is also a miniaturized liquid-phase technique performed in a 96 well-plate format. In PALME, the aqueous sample is separated from the aqueous acceptor phase by an artificial liquid membrane, created by supporting organic solvents in flat porous membranes. This procedure can promote an effective sample clean-up, in which low amounts of sample and organic solvents are required [56,57,58].

The application of PALME is compatible with high-throughput analysis such as dried blood spots [59]. Moreover, this technique can provide a number of advantages in bioanalysis when compared to traditional methodologies, such as protein precipitation, solid-phase extraction (SPE) and supported liquid extraction, including cost-effectiveness and low solvent consumption. The use of 96-well plate system can provide additional features including automatization and high sample throughput [60].

In a study reported in 2017 [61], PALME was applied for the determination of four phenethylamines, three cathinones, one piperazines, one aminoindane, one synthetic opioid, fluoroamphetamine (PFA) and methoxetamine in whole blood and plasma using UHPLC–MS/MS. A 96-well plate system was used with 0.45 µm pore size PVDF as support for the SLM. For plasma, an aliquot of 125 µL of sample was added with 115 µL of NaOH (40 mM) and 10 µL of the internal standard. Regarding whole blood, an aliquot of 100 µL of sample with either 50 µL of deionized water was added with 75 µL of NaOH (80 mM) and 25 µL of internal standard. A 5 µL aliquot of dodecyl acetate with 1% trioctylamine (w/w) was added to the PVDF membrane as SLM, and 50 µL of formic acid (20 mM) was used as acceptor solution. Extraction was conducted for 120 min at 900 rpm. Using the optimized conditions, lower limits of quantification (LLOQs) ranged from 0.1 to 0.8 ng/mL for plasma and from 25 to 147 ng/mL for whole blood. Matrix effects were found to be satisfactory within ± 15% for both matrices.

Designer benzodiazepines, benzodiazepines and Z-hypnotics were extracted from whole blood samples using PALME [62]. The experimental procedure was optimized and developed using a 96-well plate system for high-throughput analysis. An illustration of the process for PALME of the basic analytes is shown in Fig. 3. The donor solution consisted of mixing 100 µL of whole blood with 130 µL 50 mM phosphate buffer at pH 7.5 and 20 µL of the internal standard. The PVDF membrane was used for supporting the SLM, which was comprised of 4 µL of a 2-undecanone and dihexyl ether solvent mixture (1:1, v/v) with 1% trioctylamine (v/v). The acceptor solution consisted of 150 µL DMSO with 200 mM formic acid (75:25, v/v). After 60 min at 900 rpm, the extracts were diluted with deionized water (1:1, v/v), and 2 µL was analyzed by UHPLC–MS/MS system. LLOQs ranged from 2 to 15 ng/mL. Satisfactory recoveries (> 90%) were achieved for diclazepam, phenazepam, flubromazepam, meclonazepam and N-desmethyldiazepam, and the other compounds were extracted with efficiency higher than 52%.

Fig. 3
figure 3

Schematic illustration of basic compound extraction in a 96-well parallel artificial liquid membrane extraction (PALME) system. SLM supported liquid membrane

Full size image

Solid-based microextraction techniques

Solid-based microextraction approaches are gaining attention since they simplify the experimental steps and, consequently, the extraction time much shorter as compared to traditional solid-phase extraction (SPE) [63]. Solid-phase microextraction (SPME), microextraction by packed sorbent (MEPS) and micro-solid-phase extraction (µ-SPE) can be highlighted for the determination of NPS. Table 3 summarizes solid-based microextraction techniques for the determination of several new psychoactive substances.

Table 3 Solid-based microextraction techniques
Full size table

Solid-phase microextraction

SPME is an important microextraction technique developed by Arthur and Pawliszyn in 1990. The technique basically consists of a cylindrical solid support coated with an extraction phase (SPME fiber). The fiber is protected in a needle and the extraction phase can be exposed during the extraction procedure. The extraction can be performed in the sample headspace or by the direct immersion of the fiber into the sample solution depending on the properties of the analytes and matrices. Another mode known as membrane-protected SPME can also be used in some specific applications [64]. Different fiber coatings are commercially available including polymeric coatings such as polydimethylsiloxane (PDMS), polyacrylate and the combined coatings of PDMS/divinylbenzene (DVB) or DVB/carboxen/PDMS [65].

As a well-established microextraction technique, SPME already has automation devices created. CombiPAL autosampler by CTC Analytics (Zwingen, Switzerland) and TriPlus autosampler by Thermo Fisher Scientific (Milan, Italy) are devices used for the fiber SPME method, that enable automation by programming different sample preparation steps [66, 67]. Another device recently developed by CTC Analytics is the PAL (preparation and load solution) SPME Arrow, which combines the classical SPME fiber and stir bar sorptive extraction (SBSE), allowing increase of extraction yields when compared to the classical SPME fiber and maintaining its advantages [68].

SPME can be combined with gas or liquid chromatography, and also directly coupled to mass spectrometers. The use of gas chromatography (GC) allows for thermal desorption of the analytes directly into the injection port of the gas chromatograph [69]. In liquid chromatography (LC), the injection of the extracted analytes is carried out after a liquid desorption by immersing the SPME fiber into an organic solvent or adopting an online in-tube injection system compatible with automation for the injection [70]. The main variables that can be influenced on the extraction performance consist of type of fiber coating, extraction time, extraction temperature, amount of organic solvent added to the sample, pH, salt addition, sample dilution, derivatization, and desorption conditions [71]. Reviews focusing on the application of SPME for biomedical analysis [72], forensic [65] and analytical toxicology [73] have already been reported.

SPME is one of the most popular sorbent-based microextraction techniques. However, its use for the determination of NPS seems not many. A procedure for the analysis of phytocannabinoids and synthetic cannabinoids in oral fluid using direct immersion (DI) SPME coupled with gas chromatography–mass spectrometry (GC–MS) was developed by Anzilotti et al. [74]. The samples were centrifuged and an aliquot of 1 mL of supernatant was extracted with a 100 μm PDMS fiber for 15 min at room temperature, followed by desorption at 270 °C for 15 min. The fiber was previously conditioned at 250 °C for 30 min in the GC injector port. Extraction mode (DI or headspace), type of coating, extraction and desorption times were optimized. The LOQ for all analytes was 10 ng/mL, except for THC, which was 1 ng/mL, and the linear range varied from LOQ to 1000 ng/mL. Accuracy and precision were satisfactory with RSD lower than 15% [74].

In 2018, Alsenedi and Morrison [75] developed a methodology for the determination of cathinones and amphetamines in urine by SPME plus GC–MS. Several variables were optimized including pH, ionic strength, sample volume, type and volume of solvent, extraction and desorption times, agitation, and derivatization agent. Optimized procedure was carried out as follows: PDMS/DVB fiber coating was conditioned with a mixture of methanol:water (50:50, v/v) for 10 to 20 min; extraction occurred by DI using 1 mL of urine, 100 µL of internal standard, 0.5 g of NaCl and 100 µL of a 10% NaOH solution (pH 12.6) with agitation of 2000 rpm for 1 h; desorption was performed by direct immersion of the fiber coating into 65 µL of methanol for 10 min at 2000 rpm; afterwards, 10 µL of acidified methanol were added and the solution was evaporated to dryness under nitrogen steam. Derivatization was performed by adding 50 µL of pentafluoropropionic anhydride/ethyl acetate (2:1, v/v), followed by agitation, incubation for 10–15 min at 60 ºC and evaporation to dryness. Extracts were reconstituted with 50 µL of ethyl acetate and 1 µL was analyzed by GC–MS. LODs varied from 5 to 25 ng/mL and LLOQs varied from 25 to 100 ng/mL for synthetic cathinones. The method was linear (r2 > 0.99) from LLOQ to 2000 ng/mL. Precisions were satisfactory with RSDs lower than 15%.

Microextraction by packed sorbent

Microextraction by packed sorbent (MEPS) is a miniaturized solid-phase extraction, introduced in 2004 and developed for samples with volumes from 10 to 1000 µL [76]. In this technique, the sorbent is packed or coated inside a syringe (100–250 µL) as plug or cartridge. Unlike SPE, MEPS employs “up and down” (two directions) sample flow and the syringes can be reused several times. Conditioning, loading, washing, and eluting are performed when the plunger is pulled [76,77,78].

The sorbents used are typically the same as used in SPE, such as normal phases, reversed phases, mixed mode, and ion exchange materials. Also, MEPS is available for GC and LC or the 96 well-plate format for colorimetric techniques and immunoassays [79, 80]. Therefore, low sample and solvent volumes are required in MEPS with possibility of full automation by using LC or GC autosamplers [81]. MEPS performance can be affected by sample dilution, pH, cartridge conditioning, sample loading, sample washing and analyte elution; as well as the type of sorbents and solvents and their volumes [80]. Moreover, shorter analysis time can also be achieved when compared to SPE approaches [77].

MEPS has been frequently employed for the analysis of oral fluid as described in Table 3. GC–MS was used for the determination of synthetic cannabinoids, mephedrone, ketamine and dichloropane [82,83,84]. In 2019, Bianchi et al. [82] employed a commercial eVol® 50 µL syringe with a barrel-in-needle (BIN) comprised of C18 (Trajan Scientific and Medical, Melbourne, Australia). Most of the MEPS developments have involved manual operations; however, the commercially available syringes also are suitable for partial automation. Samples were pretreated by protein precipitation. The BIN was activated with 10 × 50 µL of methanol, followed by 5 × 50 µL loading sample cycles and 25 × 50 µL elution cycles carried out with dichloromethane/2-propanol/ammonium hydroxide (78:20:2, v/v/v) as extraction solvent. The evaluation of loading and eluting cycles was performed through a multivariate approach. LLOQ values for SCs were 0.25 µg/mL except for UR-144 which was 0.5 µg/mL. Mephedrone and ketamine exhibited LLOQ of 0.05 µg/mL. The linear range varied from LLOQ to 10 µg/mL for most of the analytes except for three (CP 47–497, CP 47-497C8 and HU-211), the LLOQ of which was 5 µg/mL [82].

A method for the determination of dichloropane in oral fluid samples has also been developed using MEPS [83]. The sorbent was conditioned with 3 × 100 µL of 2-propanol and 2 × 100 µL of deionized water. An aliquot of 100 µL of a mixture of oral fluid (90 µL) and 1 M phosphate buffer at pH 7 (10 µL) was eluted through the sorbent four times for extraction of the analyte. A washing step was performed using 4 × 100 µL of deionized water and 3 × 100 µL of air, and the elution was carried out with 50 µL of 2-propanol (10 cycles). The configuration proposed is shown in Fig. 4. Afterwards, 1 µL of the extract was analyzed by GC–MS. LOD for dichloropane was 70 ng/mL and LOQ was 200 ng/mL. Relative recoveries ranged from 90 to 110% for synthetic samples and 85 to 107% for real samples. Calibration curve was developed from 100 to 1000 µg/L with r2 = 0.999.

Fig. 4
figure 4

Proposed microextraction by packed sorbent procedure (MEPS)

Full size image

In 2020, Sorribes-Soriano [84] proposed an analytical method based on eVol® XR system and a 100 µL syringe with C18 as packed sorbent. Conditioning was performed with 2-propanol and deionized water. An aliquot of 480 µL of oral fluid, previously added with 20 µL of buffer, was loaded into the system by 5 charge/discharge cycles. Elution was carried out with 5 cycles using 50 µL of 2-propanol, then 5 µL of triphenyl phosphate at 5 mg/mL was added to the extracts. Following, 1 µL was analyzed by GC–MS. LOQs ranged from 30 to 60 ng/mL, and calibration curves ranging from 60 to 1000 ng/mL were obtained. Recoveries were satisfactory and ranged from 89 to 124%, except for MDMB-CHMCZCA (62 to 66%).

Moreover, MEPS coupled with liquid chromatography has also been explored for the determination of cathinones, opioids and cocaine-related compounds [85]. In this case, an aliquot of 300 µL of oral fluid sample was mixed with 200 µL of methanol and centrifuged. Then, 300 µL of the supernatant was mixed with 200 µL of 50 mM phosphate buffer at pH 9 and centrifuged again to complete sample pretreatment. Extraction procedure was conducted using an eVol® automated analytical syringe with a BIN comprised of 4 mg of C8 and SCX. An aliquot of 100 µL of previously treated sample was loaded for 6 cycles, followed by a washing step with 50 µL of water/methanol (90:10, v/v). The BIN was dried for 0.5 min and analytes were eluted with 90 µL of dichloromethane/2-propanol/ammonium hydroxide (78:20:2, v/v/v). Extracts were evaporated to dryness under a nitrogen stream at 40 °C, being reconstituted in 50 µL mobile phase for LC–MS/MS analysis. LOQs for all cathinones ranged from 0.5 to 1 ng/mL and determination coefficients from 0.9922 to 0.9984 with calibration curves prepared from LOQ to 500 ng/mL. Extraction efficiency for most of the cathinones ranged from 75 to 125%. No significant ion suppression or enhancement was observed with matrix effect from 84 and 117%, except for flephedrone which was 70% [85].

Moreover, MEPS has also been evaluated for the analysis of 17 SCs, 10 cathinones, two piperazines, methoxetamine, and 2C-B coupled to UHPLC–MS/MS [86]. Particularly, oral fluid samples were diluted with methanol and water for protein precipitation (PP). MEPS procedure was carried out using a C18 BIN following these steps: equilibration with 3 × 250 µL of methanol; conditioning with 3 × 250 µL of water/methanol (75:25, v/v); 5 × 200 µL sample loading; washing with 3 × 200 µL of water/methanol (90:10, v/v) and eluting with 5 × 100 µL of HCOOH 10 mM in methanol. The extract was added with 100 µL of ultrapure water and an aliquot of 2 µL was analyzed by UHPLC–MS/MS. Extraction variables such as composition of loading, eluting, and washing solutions, pH for the elution and effects of sonication and temperature for protein precipitation were evaluated. LOQ of 0.015 ng/mL for AM-1220 and WIN-55 were achieved. Suppressive matrix effects were lower than 40%, and recovery ranged from 40 to 96% for most of the analytes [86].

MEPS has also been used for the determination of three cathinones, four opioids, cocaine and two derivatives in plasma samples [87]. Pretreatment of plasma samples was performed with 300 µL of sample and 200 µL of methanol followed by centrifugation. This study involved a comprehensive analysis of seven variables using a fractional factorial design for the optimization of sample volume, sorbent drying time, washing volume, elution volume, number of extraction cycles, sorbent phase and pH. Optimized conditions were as follows: conditioning of C8/SCX sorbent with 100 µL of methanol and 100 µL of ultrapure water; loading of 300 µL of sample with 200 µL of 0.05 mM phosphate buffer at pH 8; washing with 150 µL of ultrapure water/methanol (90:10, v/v) and dried with 3 × 100 µL of air; eluting with 200 µL of dichloromethane/2-propanol/ammonium hydroxide (78:20:2, v/v/v). Extracts were evaporated to dryness and reconstituted with 100 µL of mobile phase prior to high-performance liquid chromatography–diode array detection (HPLC–DAD) analysis. LODs were 0.005 µg/mL, 0.01 µg/mL and 0.025 µg/mL for mephedrone, MDPV and methylone, respectively. LOQs ranged from 0.01 to 0.05 µg/mL and recoveries from 85 to 101%. Accuracy and precision exhibited acceptable RSD with values lower than 14%. Ion suppression was observed for all cathinones with matrix effect values of 42.4% for MDPV, 67.2% for mephedrone and 75.7% for methylone [87].

Micro-solid-phase extraction

Micro-SPE (µ-SPE) term was introduced in 2006 with the development of a micro-solid-phase extraction supported by a porous membrane as an alternative to SPME [88] to provide higher enrichment factors, low cost, shorter extraction time and low matrix effects [89]. In µ-SPE, a specific sorbent is conditioned, following the extraction and desorption into a small volume of organic solvent. Different configurations of this technique include spin column, pipette tip, membrane protected and dispersive µ-SPE [63, 90].

µ-SPE has been employed for the analysis of 14 SCs, four cathinones and two phenethylamines in whole blood [91]. An aliquot of 100 µL of the sample was added with acetonitrile/methanol (50:50, v/v) for sample pretreatment. Then, a mixture of 200 µL of the supernatant with 350 µL of ultrapure water was loaded in OMIX C18 tips. The tips were washed with 100 μL of H2O/acetonitrile (50:50, v/v) and conditioned with 100 μL of H2O/acetonitrile/methanol (80:10:10, v/v) prior to the sample addition. Extraction was performed using five cycles of loading/ejection. Afterwards, the tips were washed with 100 μL of methanol/H2O (80:20, v/v), and analytes were eluted using 100 μL of 10 mM HCOOH in methanol for three times. Finally, an aliquot of 6 μL was analyzed by LC–MS/MS. LOQs were 5 ng/mL for cathinones, and phenethylamines, and ranged from 0.25 to 2.5 ng/mL for SCs and their metabolites. Linearity was evaluated from LOQ to 100 ng/mL for SCs and from LOQ to 500 ng/mL for the other compounds, with r2 from 0.9935 to 0.9997. Precision and accuracy values were 15% or less for most of the analytes, and extraction recovery ranged from 21 to 70%.

Moreover, µ-SPE based on molecularly imprinted polymers (MIP) was used for the determination of 10 synthetic cathinones in urine by high-performance liquid chromatography (HPLC) coupled to mass spectrometry in tandem [92]. MIPs were synthesized using methcathinone and ethylone as templates and 50 mg were trapped into a cone-shaped 2 cm polypropylene membrane. Loading and elution conditions including type of solvents, pH, extraction phase type, temperature, shaking speed and time were optimized by univariate strategies. The MIP-µ-SPE device was conditioned with 5 mL of 0.1 M KH2PO4/NaOH buffer at pH 5 for 10 min. Cleaning of the device was performed with 2 mL of the same buffer used for conditioning, and the elution was performed with 2 mL of heptane/2-propanol/ammonium hydroxide (75:20:2, v/v/v) for 4 min. The extract was evaporated to dryness and reconstituted in 50 µL of ammonium acetate (2 mM) in methanol. LOQs ranged from 0.48 to 5.03 ng/mL and precision and accuracy were within the acceptable variation of ± 15%. Satisfactory results for the analyte recoveries were achieved with values varied from 92 to 102% [92].

Automated and semiautomated configurations in sample preparation approaches

As described before in this review, PALME consists of an important configuration that comprises the features of a microextraction approach combined with the high-throughput aspects of a 96-well plate system. This configuration has already been adopted to determine NPS in some biological matrices. In this regard, some novel and alternative strategies based on these automated and semiautomated configurations can be further exploited to provide new horizons in NPS determinations. A brief discussion involving recent approaches on this progress that can be adopted is described as follows.

Particularly, a microextraction procedure based on 96-well plate approach associated with thin-film SPME has been examined [93]. This configuration consisted of a series of solid supports coated with a thin layer of biosorbent (cork) to determine some emerging organic pollutants from aqueous samples with separation/determination by HPLC–DAD. In this case, a series of pins coated with the sorbent phase allowed for the simultaneous analysis of up to 96 samples. The solid biosorbent used as the extraction phase was thermally pretreated, sanded, and the powder was adhered to each pin with aid of a double-sided adhesive film. This sustainable approach exhibited interesting features and very satisfactory analytical results.

In addition, liquid-based microextraction approaches coupled to 96-well plate have been recently reported for the determination of environmental pollutants in aqueous samples [94]. In this particular case, magnetic ionic liquids were adopted as magneto-active solvents (extraction phases) using an analytical strategy named parallel single-drop microextraction (Pa-SDME). Interestingly, these neat solvents did not require any solid magnetic particle, because they exhibit paramagnetic components within the chemical structure. In this strategy, a set of magnetic pins were used to stabilize the solvent drops that can be used to simultaneously analyze up to 96 samples. This high-throughput configuration was coupled with HPLC–DAD and provided consistent and reproducible results.

Moreover, a membrane-based approach based on hollow-fiber renewable liquid membrane extraction (HFRLM) coupled to a 96-well plate has been recently proposed for the determination of cocaine and its metabolites in urine samples using liquid chromatography–electrospray ionization-quadrupole time-of-flight-mass spectrometry [95]. In this strategy, polypropylene membranes of 1 cm length were inserted in a series of pins coupled to a 96-well plate system, and their pores were filled with small amounts of organic solvent to create a supported liquid membrane (SLM). Afterwards, the set of pins was immersed in the urine samples, and 20 µL of organic solvent was added to the sample allowing for the renewable liquid membranes. This strategy provided a high-throughput sample preparation step, and it consists of an interesting alternative to be further exploited in NPS determination. An illustration of the process is shown in Fig. 5.

Fig. 5
figure 5

Schematic illustration of hollow-fiber renewable liquid membrane extraction (HFRLM) coupled to a 96-well plate system

Full size image

Another interesting approach in microextraction-based techniques consists of in-tube SPME. This technique has been applied for the determination of non-steroidal anti-inflammatory drugs in aqueous samples by LC–MS/MS [96]. In this automated on-line strategy, generally, a capillary column containing the inner surface coated with the extraction phase is placed between the injector (autosampler) and the liquid chromatographic column. Therefore, the sample can be loaded using a number of draw/eject cycles which allows for the extraction of the analytes by the inner solid phase. Afterwards, a desorption cycle using an organic solvent is generally applied to carry the analytes to the separation column. In this particular case, 15 analytes were efficiently extracted and determined using an automated in-tube SPME approach.

Significant efforts in developing fully automated experimental tools for sample preparation have been conducted. Fleischer et al. [97] proposed a flexible and fully automated dual-arm robot system for the determination of cholesterol in biliary endoprosthesis by GC–MS. This sophisticated system offers a number of advantages including various robot-type movements, fully automation and possibility of 24/7 operation. However, the entire system consists of a more sophisticated tool compared to the previously described configurations, and the costs of this system can be a limitation in a number of laboratories.

More specifically related to microextraction approaches, an automated open-source platform to integrate the sample preparation step and instrumental analysis has been recently proposed [98]. In this study, a lab-made cartesian robot was projected and evaluated for the determination of polyaromatic hydrocarbons using SDME, HF-LPME and MEPS coupled to liquid chromatography. The hardware of this prototype was carefully designed and integrated with electronic components to allow for fully automated procedures. Excellent results regarding accuracy and precision with relatively low cost were achieved by using the lab-made robot. This system can be a valuable tool to be further exploited also in NPS determinations.

Discussion and conclusions

The determinations of NPS are significant analytical challenges in toxicology and related areas. Microextraction techniques exhibit interesting features, because they enable rapid sample preparation, low solvent and sample consumption, and cost-effectiveness as highlighted in this review before. Moreover, the principles of “green analytical chemistry”, mostly focused on reduction or elimination of toxic reagents and solvents, and miniaturization.

New liquid and solid extraction phases have been an alternative to reduce the use of toxic reagents and solvents. Among these green solvents, amphiphilic solvents, ionic liquids and deep eutectic solvents (DESs) are the main types being used, while new sorbents include, for example, MIPs, metal–organic frameworks, and covalent organic frameworks. These are called green solvents and sorbents, and they can be used as an environment-friendly alternative to conventional ones in microextraction techniques [38, 39].

Regarding the determinations of NPS, MEPS was one of the most explored microextraction techniques. It presents desirable features such as volume handling on the microliter scale and the possibility of automation as well as online coupling with analytical systems [82,83,84,85,86,87]. Extraction efficiencies were higher for MEPS, as well as for other solid-based sample preparation methodologies when compared with liquid-based approaches. However, there is a limitation on commercially available sorbents, which narrows selective analysis and applications [38]. Low recovery rates presented by liquid-based microextraction techniques such as < 10% for some SCs in whole blood [48] and 2C-E in whole blood [61] can be explained by the physicochemical properties of the analytes and the matrix complexity. However, liquid-based methodologies stand out by the employment of lower sample volumes (50–500 µL).

Furthermore, trends toward fully automated extraction techniques are a hot topic in analytical chemistry and toxicology. Automation of extraction methods minimizes operational errors since it couples the sample preparation with the analytical instrument, also enabling improvements in reproducibility, sensitivity, and precision. Therefore, the possibility of automated microextraction techniques can offer advantages when performing procedures with multiple steps or difficulty to reproduce manually, together with avoiding contamination of the system, which leads us to search for automated approaches of these techniques. Moreover, microextraction procedures described in this review have reports on automation approaches, but there is yet a lack of unified operation modes and application in toxicological analysis [57, 99,100,101,102,103,104].

Finally, another trend in sample preparation approaches consists of the development of chip-based devices, a technique that remains poorly explored for drugs of abuse determination. This promising format uses a microfluidic apparatus, generally, manipulating small amounts of fluids (10–9 to 10–18 L) [105]. The proposed configuration exhibits advantages due to low consumption of reagents and solvents combined with reduced extraction time. Because of these interesting features, these devices can also be further evaluated for the determination of NPS in biological matrices. However, some limitations can also be noticed regarding the production of the chip-based extraction systems [105,106,107,108].