Introduction

Synthetic cannabinoid receptor agonists (CRAs) (Fig. 1), first identified in 2008 in herbal mixtures in Germany and Japan [1, 2] represent the fastest growing class of new psychoactive substances (NPS) in Europe, with 30 new compounds reported via the early warning system of the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) in 2012 [3] and 29 in 2013 [4]. A common mode of distribution of herbal products laced with these substances is the Internet, with online shops often offering a wide variety of “legal high” products. Similar to the increase in the number of identified CRAs, the number of online shops also continues to grow, with 693 shops identified in January 2012 by the EMCDDA [3]. Because the plant material is either sprayed with or soaked in a drug solution, a significant health risk arising from the use of these products is posed by the inhomogeneity of the herbal mixtures regarding the amount of active ingredient(s) per package and uneven distribution within each package [5–7]. Such variability in the product makes it impossible for consumers to safely dose these drugs, because two joints prepared from the same mixture could contain extremely different amounts of the active substance. Furthermore, the compositions of these herbal mixtures change rapidly over time, and a certain product name does not guarantee the same composition of compound(s) between batches [8].

Despite such an obvious risk, no investigation to date has been carried out on a large number of samples from one online shop to assess product variability. Most studies of these dubious products used a homogenization step prior to quantitation, and no intrapackage variability was assessed. Although homogenization is a necessary approach in a forensic chemical analysis, the results do not reflect the risk of these drugs for consumers who might be exposed to dangerous amounts upon consuming the product.

Fig. 1
figure 1

Chemical structures of the cannabinoid receptor agonists (CRAs) detected from the herbal mixtures and of the utilized internal standard AKB-48-5F

Full size image

In the present study, 311 herbal mixtures covering 31 different brands seized from a single online retailer were quantitatively analyzed utilizing a high-performance liquid chromatography-photodiode array detection (HPLC-DAD) method after screening by gas chromatography–mass spectrometry (GC–MS). Following the death of a person after consuming the herbal mixture “ACME” (containing JWH-210) and the “bath salt” product “9/11” (containing 4-methylethcathinone), the aim of this study was to assess the danger of overdosing due to interpackage and intrapackage inhomogeneities of CRAs in these products.

Materials and methods

Chemicals and reagents

Formic acid (Rotipuran® ≥98 %) was purchased from Carl Roth (Karlsruhe, Germany); methanol (HPLC grade) from J.T.Baker (Deventer, The Netherlands); acetonitrile (ACN), ammonium formate (99.995 %), ethanol (analytical grade), and ethyl acetate (analytical grade) from Sigma Aldrich (Steinheim, Germany). Deionized water was prepared using a cartridge deionizer from Memtech (Moorenweis, Germany). RCS-4 [(4-methoxyphenyl)(1-pentyl-1H-indol-3-yl)methanone] was purchased from Cayman Chemical (Ann Arbor, MI, USA). AKB-48-5F, UR-144, and XLR-11 were kindly provided by Ilmari Szilvay (Finnish Customs Laboratory). AM-1220 [(1-[(1-methyl-2-piperidinyl)methyl]-1H-indol-3-yl)-1-naphthalenyl-methanone] was obtained by purification of a research chemical using thin-layer chromatography (TLC) [9], and MAM-2201 ([1-(5-fluoropentyl)-1H-indol-3-yl](4-methyl-1-naphthalenyl)-methanone) was extracted from a herbal mixture [10, 11]. All other synthetic cannabinoids, JWH-081 [(4-methoxy-1-naphthalenyl)(1-pentyl-1H-indol-3-yl)-methanone], JWH-122 [(4-methyl-1-naphthalenyl)(1-pentyl-1H-indol-3-yl)-methanone], JWH-203 [2-(2-chlorophenyl)-1-(1-pentyl-1H-indol-3-yl)-ethanone], JWH-210 [(4-ethyl-1-naphthalenyl)(1-pentyl-1H-indol-3-yl)-methanone], JWH-307 ([5-(2-fluorophenyl)-1-pentyl-1H-pyrrol-3-yl](naphthalen-1-yl)methanone), AM-2201 ([1-(5-fluoropentyl)-1H-indol-3-yl]-1-naphthalenyl-methanone), and AM-2232 [3-(1-naphthalenylcarbonyl)-1H-indole-1-pentanenitrile] were provided by the German Federal Criminal Police Office (BKA), the State Bureaus of Criminal Investigation (LKA) Baden-Württemberg and Niedersachsen, or purchased as “research chemicals” over the Internet. 1H and 13C nuclear magnetic resonance (NMR) spectroscopy, GC–MS, and TLC were used to verify the identity and purity (≥98 %) of substances not obtained as certified standards.

Samples

All investigated herbal mixtures investigated were part of a single seizure conducted in March 2012, covering a total of 4,127 packages (31 different brands) from an online shop selling “legal highs”. From this seizure, 311 packages were quantitated as an adequate sample for interpretation of the homogeneity. Furthermore, 34 packages (21 different brands) were completely analyzed (in 200-mg portions) to investigate intrapackage inhomogeneities.

Extraction of synthetic cannabinoids

Two hundred milligrams of each herbal mixture was accurately weighed into a test tube and extracted three times by addition of 2 ml of methanol and ultrasonication for 15 min; the extracts were combined and filtered through a 0.22-µm filter (Carl Roth, Karlsruhe, Germany). The filtrate was used for in-house screening by GC–MS and quantitation by HPLC-DAD as described below. In order to mimic the usual conditions of actual drug use, no homogenization was carried out prior to sampling.

Quantitation of synthetic cannabinoid receptor agonists

For quantitation, 5 µl of filtrate was transferred into an HPLC vial containing 1 ml of mobile phase B as well as 25 µg/ml of CRA AKB-48-5F as internal standard (IS) and analyzed by HPLC-DAD utilizing a Dionex UltiMate 3000 RSLC HPLC system (Thermo Fisher Scientific, Dreieich, Germany). The system consisted of an HPG-3400RS binary pump, an SRD-3600 solvent rack degasser, a WPS-3000TRS autosampler, a TCC-3000RS column compartment, and a DAD-3000RS diode array detector (DAD). Separation was carried out by a method similar to that described by Huppertz et al. [12]. Mobile phase A consisted of deionized water with 1 % ACN, 2 mM ammonium formate, and 0.1 % formic acid. Mobile phase B was ACN containing 2 mM ammonium formate and 0.1 % formic acid. Gradient elution was performed on a Kinetex C18 column (100 × 2.1 mm i.d., 1.7 µm particle size) with a corresponding guard column (both columns from Phenomenex, Aschaffenburg, Germany). Gradient elution was started with 20 % mobile phase B for 1 min, increased to 60 % B within 1.5 min, increased to 65 % B within another 1.5 min, held for 1.5 min, increased to 99 % B within 2.5 min, and held for 2 min. Starting conditions were restored within 0.2 min, and the system was allowed to re-equilibrate for 1.8 min prior to injection of the next sample. The flow rate was set to 0.5 ml/min and the injection volume was 2 µl. The autosampler tray temperature was 10 °C and the column oven temperature was 40 °C. Wavelengths of the DAD were set to 209 and 217 nm for analysis of CRAs according to the United Nations Office on Drugs and Crime manual [13]. Five-point calibration curves (1–50 µg/ml) were prepared by fortifying 1 ml of mobile phase B containing 25 µg/ml of IS (AKB-48-5F) with the corresponding amounts of CRAs. An HPLC-DAD chromatogram of one extract from a herbal mixture containing two CRAs is shown in Fig. 2.

Fig. 2
figure 2

High-performance liquid chromatography-photodiode array detection chromatogram of one extract from the herbal mixture “Manga Hot” containing 52 mg/g AM-2201 (peak 1) and 73 mg/g JWH-210 (peak 5). Peaks 2 and 3 are impurities from the HPLC system that were also present in blank samples (not shown). Peak 4 represents the internal standard AKB-48-5F

Full size image

Verification of synthetic cannabinoid receptor agonists

Considering that many CRAs have similar absorption maxima, one filtrate of each CRA composition detected in the products was verified using GC–MS analysis before quantitation by HPLC-DAD. For this purpose, 4 µl of filtrate was transferred into a GC vial and evaporated to dryness under a gentle stream of nitrogen. The dry residue was reconstituted in 100 µl of ethyl acetate, and a 1-µl aliquot was injected into the GC–MS system, which consisted of a 6890 series GC with a 5873 mass selective detector, a 7638 B series injector, and used Chemstation G1701GA version D.03.00.611 software (Agilent, Waldbronn, Germany). GC–MS conditions were: injection, splitless mode; injection port temperature, 270 °C; column, HP-5-MS capillary (30 m × 0.25 mm i.d., 0.25 µm film thickness; Agilent); carrier gas, helium; flow rate, 1 ml/min; oven temperature, 100 °C for 3 min, ramped to 310 °C at 30 °C/min, and held at 310 °C for 10 min; transfer line heater and ion source temperatures, 280 and 230 °C, respectively; ionization energy, 70 eV in electron impact ionization mode. The obtained mass spectra were compared with those of the Cayman Spectral Library [14] and with spectra from an in-house library containing a wide range of CRAs.

Results and discussion

Interpackage inhomogeneity

Interpackage inhomogeneities in CRA content in the 29 products analyzed (27 brands) and those of the five representative brands are listed in Table 1 and in Fig. 3, respectively. The highest standard deviation (SD) detected was in the product “Summerlicious,” with the AM-2232 content ranging from 26 to 100 mg/g (n = 23). It is particularly concerning that the highest measured concentration (100 mg/g) was an outlier and could lead to severe intoxication in consumers who are accustomed to using an amount of drug material adjusted to the median CRA content. Considering the high receptor binding affinity of AM-2232 at 0.28 nM toward the CB1 receptor [15], this wide range is even more worrying. Interpackage inhomogeneities observed by Choi et al. [6] in five South Korean products (2–12 packages) were higher in most of the products investigated by them as compared to those of the present study.

Table 1 Interpackage inhomogeneity of cannabinoid receptor agonist (CRA) contents in the herbal mixtures
Full size table
Fig. 3
figure 3

Box whisker plots of the CRA contents in five different brands of herbal mixtures [MNK n = 55; Monkees Goes Bananas (M.G.B.) n = 33; Summerlicious n = 23; OMG n = 22; ACME n = 19]. Small open circles indicate outliers

Full size image

Intrapackage inhomogeneity

The intrapackage inhomogeneity of the 34 packages (21 brands) analyzed is shown in Table 2, and intrapackage inhomogeneity of the five packages of the herbal mixture ACME is also shown in Fig. 4. Surprisingly, the highest standard deviation of 20 % was detected in a package of the product with the lowest amount of herbal material, “MNK”. This product contains JWH-210, a CRA with a very high binding affinity toward the CB1 receptor (0.46 ± 0.03 nM) [16].

Table 2 Intrapackage inhomogeneities of CRA contents in the herbal mixtures
Full size table
Fig. 4
figure 4

Box whisker plots showing intrapackage inhomogeneity of JWH-210 content in five different packages of the herbal brand “ACME”. All packages were analyzed completely in portions of 200 mg of herbal mixture leading to 10 samples per package (n = 9 for package No. 3)

Full size image

Uniqueness of brand names

In 5 of the 31 brands, the CRA composition was not identical in the packages tested (Table 3). Comparing the binding affinities of the different CRAs (Table 3) detected in a particular brand, such variability can pose severe health risks to consumers. As an example, for the herbal mixture “Blaze,” one package contained JWH-307, while the another contained JWH-210, a CRA with a more than tenfold higher binding affinity toward the CB1 receptor. Despite such a large difference in binding affinity, both CRAs were added by the manufacturer in a similar percentage. Similar results were found for the most of the other brands, in which the amounts of CRAs sprayed on the plant material were similar, but the CB1 receptor affinities were up to 50-fold different. Similar substitutions were also observed by Shanks et al. [8]; however, CRA contents were not quantitated and the products might have been purchased from different vendors.

Table 3 Herbal mixtures marketed under identical brand names with different CRA compositions together with the binding affinities of the respective CRAs for the CB1 receptor
Full size table

Conclusions

Based on the obtained results, we conclude that it is impossible for users of dubious herbal mixtures to accurately dose the drugs they contain, and accidental overdoses are likely to occur frequently. Another aggravating factor is the intrapackage inhomgeneity of up to 20 % (SD) in some of the products. A third major health risk is the substitution of CRAs in herbal mixtures without changing the brand name. In almost all of these cases observed; there was a pronounced difference in the binding affinities of the respective CRA without any noticeable change in the amount added to the plant material.