Determination of 37 fentanyl analogues and novel synthetic opioids in hair by UHPLC-MS/MS and its application to authentic cases

The recent emergence of new fentanyl analogues and synthetic opioids on the drug market poses a global public health threat. However, these compounds cannot typically be identified using existing analytical methods. In this study, we aimed to develop and validate a rapid and sensitive method based on ultra-high-performance liquid chromatography–tandem mass spectrometry (UHPLC-MS/MS) for the simultaneous determination of 37 fentanyl analogues and novel synthetic opioids in hair samples. Hair samples (20 mg) were extracted by cryogenic grinding in an extraction medium of methanol, acetonitrile, and 2 mmol/L ammonium acetate (pH 5.3). Following centrifugation of the samples, the analytes were separated using a WATERS Acquity UPLC HSS T3 column. The limits of detection (LODs) and limits of quantification (LOQs) ranged from 0.5 to 2.5 pg/mg and from 2 to 5 pg/mg, respectively. The intraday and interday precisions were within 13.32% at LOQ, low, medium, and high levels. The accuracies were within the range of 85.63–116.1%. The extraction recoveries were in the range of 89.42–119.68%, and the matrix effects were within the range of 44.81–119.77%. Furthermore, the method was successfully applied to the detection and quantification of fentanyl and sufentanil in hair samples from two authentic cases. Thus, this method has great potential for detecting fentanyl analogues and novel synthetic opioids in forensic work.

Hair specimens. Blank hair samples, provided voluntarily by the laboratory staff, were used for spiked calibration standards and quality control (QC) samples. The real hair samples used were from suspected users who were arrested and investigated by police. All the samples were stored at room temperature until analysis. All participants provided written informed consent and all study protocols were approved by the Ethics Committee of Academy of Forensic Science, Shanghai, China.

Sample extraction.
To remove contaminants, the samples were washed with water two times and acetone three times, and then air-dried at room temperature. The washed samples were cut into 2-3 mm pieces and weighed (20 mg) in 2 mL tubes. Then, ceramic beads and 1 mL of the EM (containing 10 ng/mL IS) were added to tube. Subsequently, the hair samples were extracted by cryogenic grinding using a Bead Ruptor system (OMNI, Kennesaw, GA, USA) at a speed of 6 m/s for 20 s and then allowed to cool for 40 s. This process was repeated 10 times. The pulverized samples were centrifuged for 3 min at 14,000 × g and filtered (pore size 0.22 μm). Finally, 200 μL of filtrate was transferred into an autosampler vial.
UHPLC-MS/MS conditions. The UHPLC-MS/MS analysis was performed on an Acquity UPLC system (Milford, MA, WATERS, USA) coupled to a QTRAP 6500 PLUS triple quadruple linear ion trap mass spectrometer (AB SCIEX, Framingham, MA, USA). Sample separation was performed using on a WATERS Acquity UPLC HSS T 3 column (100 mm × 2.1 mm, 1.8 μm) fitted with a 1.8 μm HSS T 3 guard column. The mobile phase was composed of 20 mmol/L ammonium acetate solution containing 0.1% formic acid (mobile phase A) and acetonitrile (mobile phase B). The temperature of the autosampler was set at 4 °C and the injected volume was 5 μL. The gradient elution procedure is shown in Table 1.
The mass spectrometer was equipped with an electrospray interface operating in positive ionization mode. The source temperature and ion spray voltage were set to 500 °C and 5,500 V, respectively. The gas parameters were set as follows: collision-activated dissociation (CAD) gas, medium; curtain gas (CUR), 30 psi; nebulizing gas, 40 psi; and heater gas, 40 psi. Detection was performed using multiple reaction monitoring (MRM) with two transitions for each analyte and IS. The first transition was used for quantification and the second for qualification. The MRM transitions and optimized mass spectrometric parameters for each compound are listed in Table 2.
Method validation. The method was developed according to the Society of Hair Testing (SoHT) 33 guidelines and several recent criteria [34][35][36] for method validation. Furthermore, the recovery and matrix effect (ME) were evaluated as described by Matuszewski et al. 37 .
Selectivity. The method selectivity was assessed using eight different sources of blank hair and spiking with the IS (10 ng/mL) to evaluate potential interference. Moreover, interference from possible coadministered medications was investigated according to our previous procedure 38 .

Limits of detections (LODs) and limits of quantification (LOQs).
To determine the LODs and LOQs, blank hair samples were spiked with analyte concentrations of 5.0, 2.5, 1.0, and 0.5 pg/mg, and three replicates of each concentration were analyzed. The concentration that gave a signal-to-noise (S/N) ratio greater than 3 for both the MRM transitions was chosen as the LOD. The LOQ was defined as the lowest calibration point with a coefficient of variation (CV) of less than 20% for the precision and accuracy in the range of 80-120%.
Calibration standards (2.0-2,500 pg/mg) were prepared by adding the working solutions to 20 mg of blank hair. In addition, QC samples were prepared at four concentration levels: LOQ (2 and 5 pg/mg), low (10 pg/mg), medium (500 pg/mg), and high (2,000 pg/mg). To determine linearity, seven sets of calibrators (two replicates for each set) were analyzed. The calibration curves were constructed by plotting the peak area ratios between each analyte and IS versus the concentration using 1/x weighting.
Accuracy and precision. The method precision and accuracy were assessed by analyzing spiked blank hair samples at four QC levels (LOQ, low, medium, and high). The precision was expressed as the CV. The intraday and interday precision were determined by analyzing six replicates on one day (n = 6) and over four days (n = 24), respectively. The CV for the precision should not exceed 15% for the low, medium, and high samples, whereas that for the LOQ sample should not exceed 20%. The accuracy was determined as the percentage ratio of the measured and theoretical values.
Recovery and ME. According to the method recommended by Matuszewski et al. 37 , the extraction recovery and ME were assessed at low (10 pg/mg), medium (500 pg/mg), and high (2000 pg/mg) levels. Hair samples from six drug-free individuals were used. For each level, the samples were divided into three groups (sets 1, 2, and 3). Set 1 consisted of neat standard solutions containing all the analytes in the EM. Set 2 was obtained by extracting the blank hair samples of six individuals and then spiking with the analytes. Set 3 was obtained by extracting the spiked hair samples using the method described in Sect. Sample extraction. The extraction recovery was calculated as the percentage ratio of the peak area of set 3 to the peak area of the set 2. The ME was defined as the percentage ratio of the peak area of set 2 to the peak area of set 1.
Stability. The stability of each analyte in hair was determined by injecting the extracted samples at three levels (n = 6) after storage in the autosampler at 4 °C for 24 h.
Ethics approval and consent to participate. The hair collection was carried out in accordance with SoHT guidelines. All participants provided written informed consent and all study protocols were approved by the Ethics Committee of Academy of Forensic Science, Shanghai, China.

Results and discussion
Method development. Chromatographic conditions. In general, screening methods for fentanyl analogues requires must address the separation of isomers, e.g., PFBF and 4-fluoroisobutyryl fentanyl. Fogarty et al. 39 has reported a method for detecting 18 fentanyl analogues in whole blood and separating three pair of isomers (butyryl fentanyl and isobutyryl fentanyl, para-fluorofentanyl and ortho-fluorofentanyl, and β-methylfentanyl and α-methylfentanyl). In our study, 31 fentanyl analogues including 5 pairs of isomers were analyzed. To separate the isomers, we optimized the gradient elution based on previous studies 13 . First, we compared analyte separation using a WATERS T 3 column and a RESTEK PPFP column (100 × 2.1 mm, 5 μm) and found that better separation was achieved using the former column ( Fig. 1a).Furthermore, as the method for isomer separation in the previous study took a long time (30 min), the T 3 column was still used 13 . Based on other previous reports 39, 40 , we used methanol instead of acetonitrile and found that the solvent did not influence the separation of the chromatographic peaks significantly (Fig. 1b). With the previous method 13 , the gradient elution time was extended to achieve isomer separation. As shown in Fig. 1c, this method cannot separate the isomers. Following refinement of the gradient, the separation was still not ideal as shown in Fig. 1d. Therefore, we reduced the flow rate from 0.3 to 0.2 mL/min, which allowed separation of four pairs of isomers, but not para-fluorofentanyl and ortho-fluorofentanyl (Fig. 1e).
Sample extraction conditions. The sample preparation conditions were also optimized. Methanol is typically selected as the extraction solvent for hair samples in previously methods for quantifying fentanyl analogues 3,41,42 . However, the EM has also been used to extract analytes from hair samples 32 . Hence, methanol and the EM were compared as extraction solvents in our study. The chromatographic behavior of the compounds was better when EM was used as the extraction solvent, especially for the isomers. Subsequently, extraction using different volumes (500, 800, and 1,000 μL) of the EM was investigated. The recoveries of all compounds were in the range of 84.34-96.08%, 89.10-108.63%, and 83.27-104.15% with volumes of 500, 800, and 1,000 µL, respectively, with ME values in the range of 358.40-504.59%, 79.49-115.06%, and 77.92-104.30%, respectively. The EM volume had a great impact on the ME. In particular, when the analytes were extracted with 500 μL of the EM, the ME value increased significantly. However, for the recovery, the effect of the EM volume was not significant. Finally, the extraction solution was EM and the volume was 1,000 μL.  Table 2 and the chromatograms of all the analytes in hair samples spiked at the LOQ concentration are shown in Fig. 2.
Linearity, LOD, and LOQ. Table 3 summarizes the LOD, LOQ, regression equation, and R 2 value obtained for each analyte. To the best of our knowledge, this is the first method for quantifying 31 fentanyl analytes and 6 novel synthetic opioids. The LODs for all the compounds ranged from 0.5 to 2.5 pg/mg, and the LOQs ranged from 2 to 5 pg/mg. Busardò et al. 25 reported a method to quantify 22 fentanyl analogues in hair with LODs of 3-7 pg/g and LOQs of 11-21 pg/g. The calibration curves of all the analytes were established in different concentration range, but with acceptable correlation coefficients (R 2 > 0.99). According to previous reports 3, 25,41,43,44 , the concentrations of fentanyl analogues in hair samples are typically in the following ranges: 3-2,800 pg/mg for fentanyl, 15.1-149 pg/mg for norfentanyl, and 3-104 pg/mg for 4-ANPP. In addition, a concentration of 44 pg/ mg has been reported for furanyl fentanyl 41 . Therefore, the linearity ranges obtained for these compounds in our study cover the ranges observed in authentic cases.
Precision and accuracy. The precision and accuracy obtained for each analyte are listed in Table 4. The intraday and interday precisions of the all compounds at the LOQ, low, medium, and high levels were less than 13.32%. Furthermore, the accuracies at the four levels ranged from 85.63% to 116.1%, except for acetyl norfentanyl at the LOQ, which had an accuracy of 116.1%. Thus, the precision and accuracy of the method are acceptable according to the previous criteria [34][35][36] .
Recovery and ME. The extraction recovery and ME data are summarized in Table 5. The recoveries of all the analytes from the QC samples at four levels ranged from 89.42 to 119.68%. The ME values were within the range Stability. The stability results for each analyte are shown in Table 6. The stabilities at the three concentration levels were within the range of 77.44-113.71% for all the analytes after storage in the autosampler at 4 °C for 24 h. Therefore, the developed method is suitable for use in daily forensic work.
Application to authentic cases. Following validation, the developed method was applied to the determination of fentanyl and its analogues in hair from authentic cases. The MRM chromatograms of cases 1 and 2 are shown in Fig. 3. Table 3. LODs, LOQs, and linearity for analytes in hair. Anesthesia induction was performed by endotracheal intubation during surgery. One month after the operation, the patient's hair was collected and then the 0-3 cm segment of the hair sample was analyzed. In this case, fentanyl was detected at a concentration of 8.02 pg/mg. Schneider et al. 44 .reported a case in which a patient with a chronic and heavy toothache was treated with a fentanyl patch for 22 consecutive days. For a 5 cm hair sample cut into segments 0-1 cm, 1-2 cm, 3-4 cm, and 4-5 cm, the fentanyl concentration in all the segments was in the range of 60 pg/mg (LOQ) to 480 pg/mg. Compared with multiple doses, the concentration of fentanyl in hair was lower after a single dose, which may be why norfentanyl was not detected in our real case. A 51-year-old man was reported to police by his colleague. According to the informant, the man may have been using drugs for a long time. After hair collection, the hair sample was cut into three segments (S1: 0-3 cm, S2: 3-6 cm, and S3: 6-9 cm). Then, these samples were analyzed using our proposed method. Sufentanil was detected in the hair sample at concentrations of 183.91, 131.68, and 31.48 pg/mg for S1, S2, and S3, respectively. However, no metabolites were detected in the hair sample owing to the parent drugs being largely incorporated inside the keratin matrix from sweat, the bloodstream, and the sebum before metabolization. In this case, the observed concentration of sufentanil in hair will provide a reference for future forensic work.

Conclusions
In this study, a sensitive, simple, rapid, and robust UHPLC-MS/MS was developed and validated for determination of 31 fentanyl analogues and 6 novel synthetic opioids in hair samples. This method covers fentanyl analogues and novel synthetic opioids that are common or new to the drug market. Furthermore, the developed method was successfully applied to authentic cases.