Liquid chromatography-tandem mass spectrometry determination of 1 synthetic cathinones and phenethylamines in influent wastewater of 2 eight European cities

The popularity of new psychoactive substances (NPS) has grown in recent years, with 35 certain NPS commonly and preferentially consumed even following the introduction of 36 preventative legislation. With the objective to improve the knowledge on the use of 37 NPS, a rapid and very sensitive method was developed for the determination of ten 38 priority NPS (N-ethylcathinone, methylenedioxypyrovalerone (MDPV), methylone, 39 butylone, methedrone, mephedrone, naphyrone, 25-C-NBOMe, 25-I-NBOMe and 25-B- 40 NBOMe) in influent wastewater. Sample clean-up and pre-concentration was made by 41 off-line solid phase extraction (SPE) with Oasis MCX cartridges. Isotopically labelled 42 internal standards were used to correct for matrix effects and potential SPE losses. 43 Following chromatographic separation on a C 18 column within 6 minutes, the 44 compounds were measured by tandem mass spectrometry in positive ionisation mode. 45 The method was optimised and validated for all compounds. Limits of quantification 46 were evaluated by spiking influent wastewater samples at 1 or 5 ng/L. An investigation 47 into the stability of these compounds in influent wastewater was also performed, 48 showing that, following acidification at pH 2, all compounds were relatively stable for 49 up to 7 days. The method was then applied to influent wastewater samples from eight 50 European countries, in which mephedrone, methylone and MDPV were detected. This 51 work reveals that although NPS use is not as extensive as for classic illicit drugs, the 52 application of a highly sensitive analytical procedure makes their detection in 53 wastewater possible. The developed analytical methodology forms the basis of a 54 subsequent model-based back-calculation of abuse rate in urban areas ( i.e. wastewater- 55 based epidemiology).


59
New psychoactive substances (NPS) are emerging narcotic or psychotropic drugs that 60 are not controlled by legislation, but which may pose a public health threat. It must be 61 noted that here, the term 'new' does not necessarily refer to new inventions but to 62 substances that have recently become available (UNODC, 2014). The use of NPS has   All water samples were acidified to pH 2 with formic acid (98%), then centrifuged for 5 142 minutes at 6000 rpm. SPE cartridges were conditioned with methanol (6mL), water 143 (3mL) and acidified water (pH 2 with formic acid, 3mL). The IWW samples (100mL) 144 were spiked with mixed surrogate ILIS to give a final in sample concentration of 20 145 ng/L, then percolated through the cartridges at ±5 mL/min. The cartridges were then 146 washed with 5 mL acidified methanol (pH 2 with formic acid) and dried under vacuum 147 for 10 minutes. The analytes were eluted with 5 mL MeOH (2% ammonia), evaporated 148 to dryness at 35ºC under nitrogen and reconstituted to 1mL with 10:90 methanol:water.

149
Analyses were performed by injecting 3 μL of the final extract in the UHPLC-MS/MS 150 system. SRM mode was used to acquire all data, with the precursor ion of each compound being 153 the protonated molecule. In general, the most abundant product ion of each target 154 analyte was used for quantification, with one or two additional product ions and their 155 ion ratios used for confirmation. Furthermore, LC retention time was compared to that 156 of reference standards (tolerance of ± 0.1 minutes). These steps are in line with the 157 SANCO guidelines for analytical quality control and validation procedures (SANCO, 158 2013), which has been utilised previously by research groups for quantification of illicit 159 drugs in wastewater (Bijlsma et al., 2009). Each compound was quantified using its 160 ILIS as a surrogate internal standard, except N-ethylcathinone and methedrone, which 161 were quantified using butylone-d3 and methylone-d3, respectively. These ILIS were 162 chosen based on their ability to correct for matrix effects as well as having similar 163 retention times and chemical structures closely related to the analyte of interest.

164
The performance of the method was evaluated in terms of linearity, limits of detection 165 and quantification, accuracy and precision.

166
The linearity was studied by analysing standard solutions in solvent in triplicate at seven 167 concentration levels ranging from 0.5-100 ng/L. Linearity was deemed satisfactory 168 when the correlation coefficient (r) was >0.99, using weighted (1/X) least squares 169 regression.

170
For limits of quantification and detection (LOQ and LOD), two different concentration 171 levels (1 and 5 ng/L) were tested so as to provide a more accurate measure. The limit of 172 quantification (LOQ) was established as the lowest level for which the method was fully 173 validated using spiked IWW samples with satisfactory accuracy based on recovery 174 experiments (between 70-120%) and precision (relative standard deviation (RSD) ≤20%). Furthermore, a minimum of two MS/MS transitions were required at the LOQ 176 level for confirmation. All recovery experiments were performed in quintuplicate.

177
The limit of detection (LOD) was estimated using blank samples spiked at the lowest 178 analyte concentration tested, based on a signal-to-noise level of three from the 179 quantification SRM chromatogram. and acidified to pH 2, filtered influent wastewater at natural pH and acidified to pH 2 183 and MilliQ water) and at three different temperatures (20ºC, 4ºC and -20ºC) for 24 184 hours, 48 hours, 7 days, and 30 days (at -20ºC only). The samples were each initially 185 spiked at a concentration level of 100ng/L. One mL of each sample was collected at the 186 time intervals outlined above and spiked with ILIS (100 ng/L) just before analysis. 190 Only parent compounds were selected for this study. This is based on the previous 191 metabolism studies performed on the selected ten as well as related NPS in human and 192 rat urine (Caspar et al., 2015;Ibáñez et al., 2016;Kamata et al., 2006;Mardal and 193 Meyer, 2014;Meyer et al., 2010;Uralets et al., 2014), which showed that in spite of the 194 number of metabolites identified, the parent compound could still be detected in urine. 195 It has also been shown that the metabolic patterns of selected phenethylamine-based 196 designer drugs show rather slow metabolism rates, with parent drugs being the primary  optimised for each compound independently using a collision energy ranging from 10-207 50eV at 10eV intervals. The most sensitive transition was typically selected for 208 quantification (Q) while an additional one (or two) were selected for confirmation (q).

209
Some compounds only had two product ions of significant intensity, meaning that only 210 two transitions (i.e. Q and q1 only) could be monitored ( Table 1)  An investigation into the filtration losses of these compounds was made in samples at 230 both natural and pH 2, using a vacuum filter (Whatman, 0.45μm mixed cellulose ester 231 membrane). It was found that for naphyrone and the 25-X-NBOMe compounds, less 232 than 5% of the compounds were recovered following filtration, while for MDPV a 233 recovery of 55-71% was found. (Table S1) With these results, filtration was not Higher recoveries were obtained when using Oasis MCX cartridges ( Figure S3) and 252 therefore these were selected for further optimisation. For the washing step, water, pure 253 MeOH and acidified MeOH (pH 2) were tested. Drying under vacuum (or not) was also 254 tested, to determine the optimal washing/drying steps ( Figure S3). 255 As seen in Figure S3, the optimal SPE procedure involved washing with 5mL acidified 256 methanol and drying for 10 minutes, with recoveries ranging from 83%-93% ( Figure   257

258
In previous methods for NPS using MCX cartridges, different percentages of ammonia 259 in methanol have been used for elution, ranging from 0.5% to 4% (Kinyua et al., 2015a;260 Reid et al., 2014;Senta et al., 2015;van Nuijs et al., 2013). A compromise of 2% was 261 utilised in this work, and showed good recoveries for all compounds. An elution volume 262 of 5mL of 2% ammonia in MeOH solution was finally chosen. This was based on the 263 comparison of the recoveries of 5mL and 10mL elution volumes, with recoveries only 264 2% less for the 5mL than the 10mL.  Table 1). Accuracy was measured by means of recovery, with all compounds 275 showing satisfactory recovery (between 70-120%), and precision (all within RSD 20%). methods is commonly difficult due to the distinct criteria applied by the authors.

284
Average intensity q/Q ratios were calculated from reference standards in solvent of 285 different concentrations (Table 1). With RSD% values ≤11% for all q/Q ratios, there 286 was high confidence that the ratios are not concentration dependent. These were 287 compared to those experimentally obtained from sample extracts spiked at the lowest 288 level validated. The aim of this was to test the robustness of these values and to check 289 for potential matrix interferences that could affect the q/Q ratios and consequently, the 290 confirmation process. As Table 2 shows, average q/Q deviations were all below 30%

312
Additional data pertaining to the use of ILIS is shown in Table 2, for SPE recovery and 313 matrix effects, showing how well the ILIS are able to correct for matrix interferences.

315
Stability of analytes is a matter of concern in any analysis, particularly when dealing Milli-Q water) can be found in the supporting information (Figures S4-S7). These

358
In every sequence of analysis, water samples were injected between two calibration 359 curves. Two quality control samples (QCs), i.e. a blank water sample (previously 360 analyzed) spiked at LOQ and 10×LOQ levels, were also analysed. QC recoveries were 361 considered satisfactory when they were in the range of 70-120% for each analyte. Table 3 of this compound, the ion ratios were still checked. The q1 transition was found within 396 the q/Q ratio threshold (30%). Although the q2 ratio was more than 100% lower than 397 expected, if the sample was not well checked, it could have been erroneously assigned 398 as a false positive on the basis of one q/Q ratio accomplishment. 399 Figure S8 also shows the curious case of butylone, which seemed to have been detected 400 in an IWW sample from Bristol. As this figure shows, all three transitions can clearly be 401 seen, albeit at a retention time 0.17 minutes lower than in the quality control. Although 402 one q/Q ratio is within the ±30% threshold, the fact that the retention time is greater 403 than 0.10 minutes from the standard is enough evidence to say that this is not butylone.

404
However, with all the same transitions present, there is a possibility that this compound 405 could be related to butylone. Due to the retention time being less than the reference standard, this compound could be a metabolite, which, due to its more polar 407 characteristics would lead it to elute earlier. Alternatively, it could be a related 408 cathinone-derivative, for example ethylone, which has the same fragmentation, and has 409 been found in previous European samples ( Kinyua et al., 2015a). However, more 410 research is needed to confirm its identity.       N-ethylcathinone  --------MDPV -----/3 -/2 -6/-Mephedrone 61/106 d/--