The development of a suitable analytical method for the determination of the investigated psychotropic substances considered the simultaneous presence of both acidic (e.g., THCA) and basic (e.g., cocaine, MDMA, etc.) substances, highly polar substances such as GBL and/or GHB sodium salt (whose direct determination would have been impossible by GC-MS), and lipophilic substances such as THC. It follows that the final chromatographic and mass spectrometric conditions obtained were the result of the best trade-off among method sensitivity, chromatographic resolution, and matrix effect.

Both methanol and acetonitrile (ACN) were tested in a mixture with water; at the same elution gradient, it was observed that ACN reduced the intensities of all chromatographic peaks by about 1/10 compared with the use of methanol, which, however, reduced the S/N of the qualifier transition of GBL.

The mobile phase was tested by adding either formic acid or acetic acid in concentrations up to 0.1%. Acetic acid was preferred because of the most satisfactory S/N of each chromatographic peak. The highest intensities of all analyte peaks, without compromising the decreased signal of the acidic species THCA ionized in NI mode, were achieved with an addition of 0.02% of acetic acid. The addition of even small concentrations (1–2 mM) of either ammonium formate or ammonium acetate enhanced the signal intensity of most analytes but inhibited the ionization of both GHB and GBL; therefore, they could not be used in the mobile phase.

The chromatographic separation was optimized by thoroughly adjusting the UHPLC parameters such as flow rate, symmetric peaks, and autosampler and column temperatures. To avoid possible degradation of the more labile analytes (e.g., trans-∆9-THC and THCA) during the analysis batch, the autosampler temperature was always kept at 4 °C.

Optimization of the MS/MS method was performed for all the target compounds using the standard solution mixture at a concentration of 100 ng/mL. The precursor ion of every compound was selected as the most abundant peak of each spectrum. Different collision energies (from 5 to 50 V) were tested to perform the fragmentation. For each compound, the collision energies that yielded the one precursor–three products ion transitions with the best signal intensity were chosen (except for GBL and GHB-D6 for which only two transitions were selected) to build the SRM acquisition method, addressing the requirements of the 2021/808/EC [25]. The most intense transition was chosen as the quantifier, and the second and third most intense ones were chosen as the qualifiers. The SRM transitions, the optimized compound-dependent parameters such as Q1 and Q3 voltage potentials, collision energies (CE), and the related deuterated IS are reported in Table S1. All the analytes were ionized in PI mode, with the only exception being THCA, which ionized in the NI mode. The SRM transitions of each analyte were monitored only in defined time windows of 60 s centered on the related retention time. The total cycle time was 0.424 s, including a polarity switching of 5 ms and 23 events of 0.018 s.

A chromatographic separation of the target compounds and their ISs is illustrated in Figure 1. It is important to note the excellent profile of the peaks of all the analytes in a very short elution time (lower than 5 min).

A fit-for-purpose method validation was performed based on the standard practices for method validation in drug analysis. The method was validated considering LODs, LOQs, linearity range, accuracy, repeatability, reproducibility, recovery, and matrix effects.

For all analyzed samples, the active ingredient quantification was based on an eight-point calibration curve (LOQ, 0.1, 0.5, 1.0, 10.0, 50.0, 100, 250 ng/mL) using mixture standard solutions freshly prepared every week and added with their corresponding ISs (Table S1) at a concentration of 10 ng/mL. Internal calibration models with weighting factor 1/x were calculated, using the peak area ratio of the “quantifier” transition signal and the corresponding IS as the dependent variable y and the ratio of the standard over the IS concentration as the independent variable x. The quality of linear regression models was high, with R² always greater than 0.9945 (Table 1). Accuracy at the eight concentration levels was always within the acceptable range of 85.6–108.7%.

The limits of detection (LOD) and quantitation (LOQ) were calculated according to the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) [26] as the analyte concentration giving a signal of 3.3 s_B/b and 10 s_B/b, respectively, s_B being the blank standard deviation equal to the residual standard deviation (s_y/x), and b the calibration model slope. The obtained LODs and LOQs were at low-ppb levels, as shown in Table 1, and their values were more than satisfactory, considering the high concentrations of active ingredients in seized drug samples.

The obtained LODs and LOQs were lower than those reported in the literature for similar applications with respect to the investigated compounds (Table 2). In addition, the analysis time was among the lowest of those reported in Table 2.

Carry-over and memory effects were carefully evaluated. For this purpose, blank sample injections of the mobile phase after the injection of the standard solution mixture at the highest concentration level of the calibration plot were acquired. The presence of the peak of cocaine at the concentration close to the respective LOQ value was observed. Carry-over was overcome by using a proper washing solution for the autosampler, which consisted of a mixture of MeOH/isopropanol/water (80/10/10 (v/v/v)) used to clean both the needle and the pump port before and after each injection for 5 s. In addition, no memory effect was found thanks to the above-mentioned autosampler washing step and the washing of the chromatographic column with 100% of organic solvent (B) for 2 min at an increased flow-rate of 0.65 mL/min. This allowed to the chromatographic column to be fully cleaned of any impurities or residues that may have been present in the injected samples. Selectivity was investigated by the analysis of 10 consecutive injections of mobile phase (blank) samples. No interfering species were observed at the retention time of the analytes or IS in the analyzed blank samples.

In order to use the method in routine analyses, a specific control step must be applied to verify the method stability over time [26]. For this purpose, a quality control (QC) solution at 25.0 ng/mL was analyzed every 20 injections of each sample batch as part of the quality control process. All the results obtained for the QC solutions lay within the ±σ control limits of the calibration plots. Intra-day (i.e., repeatability) and inter-day (i.e., intermediate) precisions on concentration and retention time were determined by analyzing 5 replicated QC samples per day (n = 5) and for 7 days (n = 35), respectively. The obtained intra- and inter-day precisions in terms of relative standard deviation (RSD%) are reported in Table 1. They were lower than 4.1% and 12.6%, respectively, for the concentration and lower than 0.6% and 1.9%, respectively, for the retention time.

The evaluation of matrix effects is a crucial step in the analysis of challenging samples, especially if the erroneous quantification of analytes can lead to a wrong decision in judicial processes [27,28,29,30]. The addition of isotopic standards compensates for the possible presence of matrix effects, and the quantification was performed by internal calibration. As the matrices of the analyzed samples had very diverse nature (e.g., powders, agglomerates, inflorescences, oils, tobacco, cannabis resins, liquids, etc.), the labeled standards of the various psychotropic substances were used to calculate the recovery by fortifying them within the different matrices subjected to the extraction process (paragraph 2). Therefore, samples of known composition were fortified with 50.0, 250.0, and 500.0 µL of the respective labeled standard at 1000 mg/L to determine recovery directly on the matrix by the external calibration model, which was calculated in the same range of the internal calibration model. Each extraction was repeated three times.

Recovery values obtained for each analyte spanned from 94.3 ± 0.5% to 109.1 ± 0.7%, demonstrating the suitability of the developed extraction procedure even in the presence of complex matrices such as marijuana and hashish. Both types of samples gave the same recovery value as THC and THCA. The recovery values were reproducible and independent of the analyte concentration within the explored concentration range. According to the t-test (95% confidence level), there was no significant difference among the three obtained recovery values; therefore, the average recovery is reported in Table 1.