Comprehensive review of the detection methods for synthetic cannabinoids and cathinones

The Duquenois–Levine color test, which is used to identify classical cannabinoids such as Δ⁹-tetrahydrocannabinol, is negative for the synthetic cannabimimetics. The van Urk color test, which is used to identify indole-containing drugs of abuse, is also negative for these compounds. The use of 2,4-dinitrophenylhydrazine, which reacts with a keto moiety, is capable of reacting with synthetic cannabimimetics, such as the naphthoylindole, phenylacetylindole, benzoylindole, and cyclopropylindole classes, either in powder form or adsorbed onto plant material, and a positive test solution turns from yellow to orange. Although the LOD concentration was not detailed in the article, the solution tested contained at least 10 mg of cannabimimetic powder suspended in methanol (1 ml) [11]. The Marquis reagent, which reacts with all nitrogen-containing drugs, is positive for cyclohexylphenols and the JWH series. Although Dragendorff reagent is also positive for the JWH series, its LOD concentration is higher than that of Marquis reagent. Fast blue BB reacts with cyclohexylphenols, and the LOD concentration is not lower than that of Marquis reagent [12]. Iodoplatinate is also used as a detection reagent after TLC [13]. Although it is possible to detect synthetic cannabinoids with each reagent in these screening tests, it is difficult to detect small amounts or mixtures of synthetic cannabinoids.

ELISAs developed in-house could be calibrated at 5 ng/ml with the 5-OH and 4-OH metabolites of JWH-018 and JWH-250, respectively, and evaluated for the detection of synthetic cannabinoids in urine [14]. Recently, some commercially available immunoassay kits, such as DrugCheck K2/Spice Test, DrugSmart Cassette, and RapiCard InstaTest, have been developed for the detection of these drugs in urine. These devices are more useful than the colorimetric methods, because they do not require special reagents or tools, and the results are obtained easily and quickly. The devices also can detect older types of synthetic cannabinoids, such as JWH-018 or JWH-073, but, unfortunately, new designer drugs such as QUPIC and AB-CHMINACA cannot be detected.

Typical mass spectra of synthetic cannabinoids are shown in Fig. 3. Molecular (M⁺) and/or fragment ions observed by full scan data acquisition of GC–MS reflect the structures of the synthetic cannabinoids [13, 15, 16]. As shown in Fig. 4, the fragmentation pathways of naphthoylindoles have been well studied for the identification of synthetic cannabinoids by GC–MS [12, 15]. Therefore, the identification of synthetic cannabinoids is facilitated by comparison of the spectra with commercial and open databases.

In naphthoylindoles, the carbonyl group fragment ions, which are caused by α-cleavage of the alkylamino group of the indole, are typically observed. In addition, [M−17]⁺ is certainly observed in naphthoylindoles. For example, fragment ions at m/z 284 and 214 are observed in JWH-018, corresponding to those of the indole moiety caused by α-cleavage of the N-pentyl of indole and naphthoyl. Fragment ions at m/z 127 and 155 are observed in JWH-018, corresponding to the naphthalene group caused by the α-cleavage of the carbonyl group. Moreover, ions at m/z 324 are observed as [M−17]⁺ (Fig. 3a). Like naphthoylindoles, fragment ions caused by α-cleavage of the alkylamino group of the indole and carbonyl groups are shown, although [M−17]⁺ is not observed for benzoylindoles. For example, fragment ions at m/z 264 and 214 are observed for RCS-4, caused by α-cleavage of N-pentyl of the indole and 4-methoxybenzoyl. The ions at m/z 127 and 155, which are caused by naphthyl and naphthoyl moieties of naphthoylindoles (Fig. 3a), and the ion at m/z 135 caused by the 4-methoxybenzoyl moiety (Fig. 3b) are useful as precursor ions for identification of these drugs by GC–MS-MS. On the other hand, the methylpiperidine moiety is bound to the nitrogen of the indole, and the ion at m/z 98 is observed as the base peak (Fig. 3d). Unlike naphthoyl and benzoyl indoles, the base peak of the fragment ion caused by the N-alkylindole 3-carbonyl moiety for phenylacetyl (Fig. 3c), cyclopropyl, or adamantyl (Fig. 3f) indoles, is only shown in each full scan spectrum. Analogs, in which the indole skeleton is changed to an indazole, such as THJ-018, have also appeared on the market. In these analogs, molecular and N-dealkylated ions are typically observed in the spectrum (Fig. 3e).

Recently, amide- or ester-type analogs bonded with an N-alkylindole or N-alkylindazole 3-carbonyl moiety have appeared on the market [17, 18]. In these analogs, the abundance of the molecular ion is low, and the fragment ion caused by the indoyl (or indazoyl) moiety is observed as a base peak (Fig. 3f–k). Although the fragment ion caused by elimination of the terminal CO–NH₂ is lower than that of the cleavage of the amide moiety in indole analogs, such as ADBICA (Fig. 3h) [19], the fragment ion caused by elimination of terminal CO–NH₂ is as intense as that of the cleavage of the amide moiety in indazole analogs, such as ADB-PINACA and AB-CHMINACA (Fig. 3j, k) [20]. The substitution of the indole skeleton with the indazole moiety, such as in THJ-018 and THJ-2201 [21], has also been observed in these analogs. In these analogs, molecular and N-dealkylated ions are typically observed in the spectrum.

For example, in the simultaneous analysis of synthetic cannabinoid species, 10 mg of ground powder of the dried leaves was extracted with 10 ml of methanol under ultrasonication for 10 min. The extracts were centrifuged for 5 min at 3,000 rpm, and the supernatants were filtered and used for GC–MS analysis. The LODs were 0.5–1.0 mg/l, and linearity was obtained at concentrations up to 100 mg/l [16]. In another article [22], herbal samples (approximately 50 mg) were put into 10-ml headspace vials, and the vials were capped with 20-mm magnetic crimp seal caps with PTFE/silicone septa. The samples were incubated at 200 °C with pulse-agitation at 250 rpm. A StableFlex carboxen/polydimethylsiloxane fiber was inserted into the headspace for 5 min for extraction. The fiber was then injected into the GC inlet for 15 min to desorb the analytes. The LOD of synthetic cannabinoid in the samples was at least 20 μg.

The tentative identification of synthetic cannabinoids appears easy, but similar mass spectra are sometimes obtained by GC–MS because regio- and ring-substituted analogs are still distributed on the market. The misidentification of these analogs arises when using only the information from the mass spectra. When tandem and high-resolution MS are used to identify the conformational isomers or regioisomers, such misidentification does not occur [23‐28]. Moreover, identification of cyclopropyl or ester analogs, such as UR-144 or QUPIC, is usually not possible because cyclopropyl analogs are heat-unstable and are easily degraded in the injection port of the GC instrument [29, 30].

Many research groups have used LC–MS-MS for determination of synthetic cannabinoids in herbs and biological samples, and some have studied the fragmentation of synthetic cannabinoids in detail [15, 31]. The probable fragmentation pathways are shown in Fig. 4. Because the protonated molecular ion is only observed by LC–MS, and the information acquired by LC–MS is lesser than that for GC–MS, it is necessary to obtain other data that reflect the chemical structures by LC–MS-MS or TOFMS. Fragment ions are observed by product ion scanning when the protonated molecular ion is used as the precursor ion. In naphthoylindole, ions at m/z 127 and 155 are generated by naphthyl and naphthoyl moieties. However, information about the indole moiety tends to be not revealed by LC–MS-MS. On the other hand, the N-alkyl moiety of a synthetic cannabinoid is mainly modified for excretion into urine as a metabolite. Therefore, LC–MS-MS is a useful methodology to search for metabolites of synthetic cannabinoids in urine.

Recently, packages containing mixtures of multiple synthetic cannabinoids have been sold commercially, even though the package ingredients have been largely unknown to both sellers and buyers. In this aspect, the LC–MS-MS screening method is helpful in some estimation of the ingredients. Kneisel and Auwärter [32] demonstrated the simultaneous detection of 30 synthetic cannabinoids in serum; the LODs and LOQs were 0.01–2.0 and 0.1–2.0 ng/ml, respectively. There are many applications for analysis of synthetic cannabinoids in urine, hair, and oral fluids [33‐37]. The typical published methods for analysis of synthetic cannabinoids in biological materials are summarized in Table 1 [38‐54]. Simple LLE is usually used for the extraction of synthetic cannabinoids from biological materials because of the high hydrophobicity of the drugs. The chromatographic conditions are generally simple and do not require a special technique; octadecyl-type columns were used as analytical columns and analyses were performed in gradient mode.

The identification of an unknown drug in a biological material without information is almost always difficult, even if analysis is carried out with LC–MS-MS. To overcome this situation, high-resolution MS or TOFMS become helpful tools for tentative estimation of parent drugs and their metabolites of synthetic cannabinoids.

To clarify the chemical structure of an unknown drug in a herbal blend product that contains more than several milligrams of the drug, GC–MS, LC–MS-MS, and high-resolution MS (or TOFMS) can be used to estimate the structure. The target compound is then purified by preparative LC or preparative TLC to obtain more than several milligrams of the compound of high purity, which is then analyzed by NMR spectroscopy [17‐21]. The detailed chemical structure can be elucidated by the above laborious instrumental analyses.

The Marquis reagent, which reacts with all nitrogen-containing drugs, is negative for cathinones, such as cathinone and mephedrone, but is positive for cathinone analogs that have a methylenedioxy moiety in each molecule. The cathinone analogs with a methylenedioxy moiety also react with the Chen reagent, which changes to orange in positive tests. Although the LOD concentration is not reported, the described test solution contained cathinone powder (at least 10 mg) suspended in methanol (1 ml) [55‐57]. The combination of Marquis, Ehrlich, Simon, Lieberman–Burehand, and Mandelin reagents is useful for the detection of cathinones in samples. Like synthetic cannabinoids, the identification of these compounds, of course, cannot be performed using these methods; moreover, the detection of small amounts or mixtures of cathinones is difficult.

Some researchers have tried to detect cathinones in urine using immunoassay technology [58, 59]. Some articles revealed false-positive results by immunoassays; for example, MDPV was cross-reactive with phencyclidine [60]. Therefore, specific detection of cathinones by a commercial immunoassay is not yet possible.

Mass spectral profiles of cathinones are very simple in the positive mode of GC–MS, because only the base peak originating from the immonium ion in each molecule is observed. The probable fragmentation pathways of cathinones are described in previous articles [61‐64] and are shown in Fig. 5. However, this phenomenon makes the identification of cathinones difficult. To help identify cathinones, other information, such as tandem mass spectrometric data, are usually used because more structural information about the molecule is obtained.

Zuba [61] introduced the systematic identification of cathinones using the mass spectra obtained. First, it should be checked whether the molecular ion is observed. The immonium ion (m/z = 16 + 14 n, n = 1, 2, 3,…) is then checked in the EI spectrum. If the immonium ion is found in the spectrum, the substance could be a straight-chained cathinone. If not, it is checked whether the ion for a pyrrolidine ring is observed (m/z = 70 + 14 n, n = 1, 2, 3,…). If this ion is found in the spectrum, the substance could be a cathinone with a pyrrolidine ring in the molecule [61]. There are various regioisomers in cathinones. To identify the cathinones, it is necessary to assign both the location and length of the bonded alkyl chain. Moreover, the ring-substituted moiety is also needed to be assigned. Zuba [61] demonstrated the following rules: the fragment ions reflecting the ring-substituted moiety are observed at m/z 77 and 105 for a nonsubstituted phenyl ring, at m/z 91 and 119 for a methylphenyl ring, and at m/z 121 and 149 for a methylenedioxyphenyl ring. Matsuta et al. [62] demonstrated the detailed analysis of MS data obtained by GC–EI-MS for identification of cathinones and specified indexing information. However, the ionization rate of the fragments in ring-substituted cathinones is remarkably weaker than that of the immonium ion. Other information obtained by TOFMS or CI-MS is helpful to delineate the molecular structure [63‐66]. The identification of the regioisomer of the fluorinated cathinones was demonstrated using CI-MS [67]. However, this phenomenon was suggested to be limited to these analogs.

The published methods for analysis of cathinones in biological materials are summarized in Table 2 [68‐79]. Simple LLE is usually used for extraction of cathinones from biological materials. The chromatographic conditions are also simple and do not usually require a special technique.

The strategy for the detection of cathinones by LC–MS-MS is almost same as that for synthetic cannabinoids; almost all methods use MRM or SRM mode for sensitive determination. The probable fragmentation pathways are shown in Fig. 5. The [M+H]⁺ ion is selected as a precursor ion, and three product ions that reflect the chemical structures of the cathinones are selected. Using this method, 30–50 drugs are monitored simultaneously in samples [80‐82]. The published methods for analysis of cathinones in biological materials are summarized in Table 3 [68, 69, 74, 75, 77, 79, 82‐98]. Simple LLE is usually used for the extraction of cathinones from biological materials. The chromatographic conditions are also simple and do not require a special technique.

In the same way as identification by GC–MS, other information obtained by TOFMS or tandem MS is needed to clarify the molecular structure. An authentic drug or library database is needed to identify the drugs. Moreover, the probable fragmentation pathways of cathinones are described in the previous articles [61, 65, 66, 99]. These data are helpful in identifying the drugs.

As described in the section on synthetic cannabinoids, we occasionally encounter a dubious product that contains more than several milligrams of an unknown cathinone-like compound. In such a case, GC–MS, LC–MS, high-resolution MS (or TOFMS), and finally NMR spectroscopy are used to clarify the detailed chemical structure of the compound.

The common method of consumption of synthetic cannabinoids is smoking, which is the same as for conventional cannabis. The maximum concentrations of synthetic cannabinoids in serum are reached in less than 10 min after smoking [38]. The drugs absorbed in the body are metabolized smoothly, and the concentrations decrease rapidly. Moreover, there is also a report that cannabinoids accumulate in the adipose tissue because of their high lipophilicity [52]. Therefore, detection of the drug from serum is usually difficult. Synthetic cannabinoids absorbed in the human body are metabolized to hydroxyl or carboxyl derivatives of the aromatic ring or N-alkyl side chain [100]. It is difficult to identify the parent drug and its metabolites in blood by GC–MS alone because the fragmentation of the metabolites is similar to that of the parent drug and analogs. Moreover, the concentration of the unchanged synthetic cannabinoids in blood is very low, and the number of metabolites that are commercially available is small. Low sensitivity is a limitation for the determination of synthetic cannabinoids in blood by GC–MS.

Although the concentration is influenced by the sampling time after drug intake and by the intake amount, concentrations of these drugs in serum were reported in the range of 0.1–190 ng/ml in poisoning cases [46]. In fatal cases, the concentrations of the drugs in blood were 0.1–199 ng/ml for JWH-018 and 0.1–68.3 ng/ml for JWH-073 [50], 12 ng/ml for AM-2201 [100], 1.1–1.5 ng/ml for 5F-PB-22 [53], and 12.4 ng/ml for MAM-2201 [52].

Unlike synthetic cannabinoids, the most common method of consumption of cathinones are insufflation (snorting) or ingestion. Inhalation, sublingual and rectal administration, and intramuscular or intravenous injection have also been reported. Unlike synthetic cannabinoids, the concentration of cathinones in blood is thought to vary because of the many modes of administration used by abusers. Only the blood concentration at one point and at several points have been quantified, and there is no report on continuous monitoring of the profile of the drug concentration in blood. The fatal concentration of the drug in blood was reported to be around 400 ng/ml [75]. The stability of cathinones in blood samples is clearly influenced by pH, as well as in the final extracts. In blood samples preserved with NaF/potassium oxalate, the measured concentrations of cathinone, methcathinone, ethcathinone, mephedrone, and flephedrone declined by ca. 30 % after 2 days of storage at 20 °C [82].

Some groups have studied the metabolic pathways of cathinones [68, 94, 101‐103]. Unlike synthetic cannabinoids, the parent cathinones are detected easily in biological materials and are selected as the target because the unchanged parent drugs are rapidly excreted in urine. Cathinones are ionized in the body, and the reabsorption rate is low in the kidney because of low hydrophobicity. The excretion profile of α-PBP and α-PVP in human urine was determined after an intravenous injection, and the elimination half-life in urine was approximately 12 h. Moreover, the excreted amount in urine was influenced by urinary pH, like a psycho-stimulant [104]. To analyze these drugs in biological materials, it is necessary to remove endogenous substances from each sample and enrich the content of the drug. As shown in Table 3, LLE is usually used for extraction of the drugs from biological materials. The quantification of the metabolites is important to predict the hazardous properties of the metabolites. However, because there are few metabolites marketed, no detailed study about their pharmacological activity or toxicity has been conducted.

In fatal cases, the concentrations of the drugs in blood were: 560–3,300 [72], 272 [105], and 60–1,120 ng/ml [106] for methylone; 1.2–22 [107], 5.1 [108], and 5.5 µg/ml [88] for mephedrone; 55.2 ng/ml for α-PBP [95]; 486 [73] and 654 ng/ml [109] for α-PVP; 180 ng/ml [98] for PV9; 170 [70], 82 [110], 1,200 [74], 440 [75], 17–38 [92], and 700 ng/ml [111] for MDPV.