All chemicals and reagents used were of analytical grade. Methanol (≥ 99.9%) and acetonitrile (≥ 99.9%) were purchased from Merck (Darmstadt, Germany); ultrapure water was obtained from Biosolve Chimie (Dieuze, France). Formic acid (≥ 98%) and ammonium formate (97%) were purchased from Merck (Darmstadt, Germany). The analytical column Kinetex Phenyl-Hexyl (50 × 4.6 mm, 2.6 μm) was obtained from Phenomenex (Torrance, USA). Analytical standards consisting of NPS (Supplementary material—Table S1) were provided by the Italian National Institute of Health. Internal standard mix, consisting of deuterated compounds, was purchased from Chromsystems Instruments and Chemicals GmbH (Munich, Germany).

Blood (3) and urine (3) samples derived from paediatric cases of suspected intoxication were obtained from the emergency department of IRCCS “Burlo Garofolo” Hospital. Furthermore, postmortem specimens of blood (6), urine (4), vitreous humor (3), synovial fluid (7), cadaveric tissues of liver (4), kidney (4), spleen (3) and cadaveric larvae (5) were obtained from forensic cases with suspected cause of death due to intoxication or overdose and collected during autopsy by forensic pathologists of the University of Trieste and School of Forensic Medicine. All the biological samples were carried to the Advanced Translational Diagnostic Laboratory and stored at – 20 °C until the analysis. Biological samples of this study were left over from routine analyses and their use for analytical validation was approved by IRCCS “Burlo Garofolo” (RC 56/22).

For whole blood samples, 5 μL of internal standard (IS) mix was spiked into 90 μL of human whole blood matrix. Samples were extracted using a protein precipitation procedure. Basically, 900 μL of methanol:acetonitrile (50:50, v/v) was added to the sample and vortexed for 1 min. Then, after sonication for 3 min and vortexing for 1 min, samples were centrifuged for 5 min at 14,100 g. The supernatant was transferred into a new tube and dried under nitrogen. The residues were reconstituted with 250 μL of methanol: water (20:80, v/v).

For urine samples, 5 μL of IS and 40 μL of β-glucuronidase enzyme, whose enzymatic hydrolysis efficiency was previously tested by Chromsystems Instruments and Chemicals GmbH (Munich, Germany), were spiked into 50 μL of urine matrix. Afterwards, samples were incubated for 2 h at 45 °C to allow enzymatic deconjugation. At the end of incubation, 100 µL of precipitant reagent was added and, after vortexing, the samples were centrifuged for 5 min at 14,100 g. To 100 µL of supernatant, 150 µL of dilution buffer was added.

For vitreous humor and synovial fluid, 5 μL of IS was spiked into 50 μL of biological matrix. Afterwards, 40 µL of dilution buffer and 100 µL of precipitant reagent were added and, after vortexing, the sample was centrifuged for 5 min at 14,100 g. To 100 µL of supernatant, 150 µL of dilution buffer was added.

For cadaveric tissues of liver and kidney, as well as for cadaveric larvae, 500 mg of matrices was weighed and subsequently 1.5 mL of methanol was added. Specimens were homogenised using the instrument Bead Ruptor Elite (Omni International, Milan, Italy) according to specific grinding protocols previously set up (liver: 2 cycles at 5 m/s for 20 s without pause; kidney: 3 cycles at 4 m/s for 10 s with a pause of 10 s; larvae: 3 cycles at 6 m/s for 15 s with a pause of 5 s). The homogenate was centrifuged for 10 min at 14,100 g. One mL of supernatant was dried under nitrogen and the residues were resuspended in 50 μL of mobile phase A. Then, sample preparation was the same as that for urine matrix.

For cadaveric tissues of spleen, 500 mg of spleen was weighed and subsequently 1.5 mL of methanol was added. Samples were homogenised as described above with the protocol: 2 cycles at 4.5 m/s for 10 s with a pause of 10 s. Then, the homogenate was centrifuged for 10 min at 14,100 g, and 150 µL of dilution buffer was added to 100 µL of supernatant.

Analyses were performed with a HPLC Exion LC 2.0 (Sciex, Milan, Italy) combined with a QTRAP 6500 + system (Sciex, Milan, Italy). To achieve chromatographic separation, gradient elution of mobile phase A (10 mM ammonium formate) and B (0.05% formic acid in methanol) on the analytical reverse-phase column Kinetex Phenyl-Hexyl (50 × 4.6 mm, 2.6 μm), thermostatted at 30 °C, was performed. The mobile phases were replaced every 2 days. A linear gradient (700 μL/min) from 10% B to 98% B in 7.0 min followed by 1.5 min of 98% B and 1.0 min of 10% B was employed. The total chromatographic run-time was 11 min. Sample injection volume is 15 μL. Quality controls at known composition (Supplementary material—Table S2) were injected before starting the analysis of a batch of samples to check the instrumental performance.

The ion source mass spectrometer parameters were as follows: curtain gas, 30 psi; collision gas, high; ion spray voltage, 5400 V for positive mode and − 5400 V for negative mode; capillary temperature, 500 (°C); ion source gas, 55 psi and collision gas, high. Acquisition method setting consists of a survey scan and an Information-Dependent Analysis (IDA) triggered scan. As survey scan, multiple reaction monitoring (MRM) mode with 751 transitions (704 in positive mode and 47 in negative mode) was selected. Compound-dependent parameters for each MRM transitions, such as precursor and product ions, declustering potential (DP), entrance potential (EP), collision energy (CE) and collision cell exit potential (CXP), were reported in Supplementary materials—Tables S3 and S4. Contrary to what was set up for the compounds in negative mode, scheduled MRM of compounds in positive mode was adopted analysing a time window of ± 25 s. Q1 and Q3 were used at unit resolution (0.6–0.8 amu at half height). The IDA intensity threshold was set to 30,000 and 1000 counts per second (cps) for positive and negative mode respectively. The two most intense MRM transitions per cycle exceeding the selected threshold were considered for the dependent enhanced product ion (EPI) scan. For further improvement of the identification of coeluted compounds, the MRM transitions, which triggered the dependent scan twice consecutively, were excluded for EPI scans for 15 s. The EPI scans were performed at a scan range of 50 to 700 amu using the dynamic fill time mode with a scan rate of 10,000 amu/s applying a CES of 35 ± 15 eV. The source and the compound dependent parameters were the same as used for the MRM mode. The MS/MS spectra obtained from the analysis were compared with the ones present in the MS/MS Forensic HR-MS/MS 2.1 library (1820 available spectra) (Sciex, Milan, Italy). Based on the present analytical workflow, the proposed method could be defined a MRM-IDA-EPI screening.

All the results obtained from the screening test underwent confirmation by detection of specific MRM for each analyte.

Data processing and analysis were performed using Analyst (version 1.5, Sciex, Milan, Italy) and SCIEX OS (version 2.0, Sciex, Milan, Italy). List of the rules for data processing is reported in Table 1.

Agreement between the screening test and the LC–MS/MS analysis was estimated on the cohort of postmortem specimens calculating the percentage of agreement (confirmed cases/total cases × 100). Furthermore, Cohen’s kappa analysis was also performed as statistical measurement to observe the agreement between the data sets, also taking into account the chance agreement. In particular, kappa (κ) scores between 0.81 and 1 represent perfect agreement, 0.61 and 0.8 substantial agreement, 0.41 and 0.6 moderate agreement and 0.1 and 0.2 slight agreement. Negative values may generally be interpreted as no agreement (McHugh 2012).