LC-QTOF-MS Identification of Major Urinary Cyclopropylfentanyl Metabolites Using Synthesized Standards.

Cyclopropylfentanyl is a fentanyl analog implicated in 78 deaths in Europe and over 100 deaths in the United States, but toxicological information including metabolism data about this drug is scarce. The aim of this study was to provide the exact structure of abundant and unique metabolites of cyclopropylfentanyl along with synthesis routes. In this study, metabolites were identified in 13 post-mortem urine samples using liquid chromatography quadrupole time-of-flight mass spectrometry (LC-QTOF-MS). Samples were analyzed with and without enzymatic hydrolysis, and seven potential metabolites were synthesized in-house to provide the identity of major metabolites. Cyclopropylfentanyl was detected in all samples, and the most abundant metabolite was norcyclopropylfentanyl (M1) that was detected in 12 out of 13 samples. Reference materials were synthesized (synthesis routes provided) to identify the exact structure of the major metabolites 4-hydroxyphenethyl cyclopropylfentanyl (M8), 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5) and 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl (M9). These metabolites are suitable urinary markers of cyclopropylfentanyl intake as they are unique and detected in a majority of hydrolyzed urine samples. Minor metabolites included two quinone metabolites (M6 and M7), not previously reported for fentanyl analogs. Interestingly, with the exception of norcyclopropylfentanyl (M1), the metabolites appeared to be between 40% and 90% conjugated in urine. In total, 11 metabolites of cyclopropylfentanyl were identified, including most metabolites previously reported after hepatocyte incubation.

Introduction

The fentanyl analog cyclopropylfentanyl, N-phenyl-N-[1-(2-phenylethyl)piperidin-4-yl] cyclopropanecarboxamide, is a potent μ-receptor agonist first discovered on the illicit market in June 2017 (1, 2). It has been involved in at least 78 deaths in Europe and over a hundred in the United States (2–4).

In general, studies on the toxicology of fentanyl analogs are scarce, and the metabolism has been studied only for a few using human liver microsomes (5, 6), human hepatocytes (5, 7–10) and/or authentic specimens (6, 9–11). Metabolite identification is the first step in establishing analytical methods that also target metabolites. Such methods can sometimes increase the window of detection (12), and ratios between metabolites and the parent compound can be used to estimate time of intake (13, 14). Also, metabolite identification might be relevant in forensic toxicology as some metabolites are active and contribute to drug toxicity (15).

A metabolism study incubating cyclopropylfentanyl with multi-donor human hepatocytes was recently published by our research group (7). The major metabolite (79%) was found to be the normetabolite, the second most abundant (16%) a metabolite monohydroxylated either on the piperidine ring or the phenethyl moiety. Minor metabolites were mainly generated by modifications on the phenethyl moiety producing a hydroxy metabolite, a piperidine N-oxide, a dihydrodiol, a hydroxy methoxy metabolite and a glucuronide (in order of decreasing abundance). However, for five out of seven identified metabolites the exact structure could not be determined. Not knowing the exact structure of major metabolites is problematic as reference materials needed for assay development are much harder to synthesize.

The aims of the present study were (i) to support the production of reference materials for cyclopropylfentanyl by identifying major urinary metabolites, (ii) to provide the exact structure for those metabolites using reference materials synthesized in-house as well as (iii) to provide a synthesis route for those reference materials.

Experimental

Case samples

A total of 13 authentic urine samples were obtained from Swedish autopsy cases where cyclopropylfentanyl was identified in femoral blood.

Chemicals and reagents

Cyclopropylfentanyl was obtained from Cayman Chemicals (Ann Arbor, MI, USA). Liquid chromatography–mass spectrometry (LC-MS) grade acetonitrile, water and formic acid used for the LC quadrupole time-of-flight MS (LC-QTOF-MS) were obtained from Fisher Scientific (Gothenburg, Sweden), while the ammonium formate (Fluka) was obtained from Sigma-Aldrich (Stockholm, Sweden). β-glucuronidase/arylsulfatase (Helix promatia, 4.5 U/mL and 14 U/mL, respectively) for urine treatment was purchased from Roche (Mannheim, Germany).

Gradient grade acetonitrile, methanol, formic acid p.a. (98%) and sodium acetate p.a. were purchased from Merck (Darmstadt, Germany). Water was produced in-house with a MilliQ Gradient 10 production unit (Millipore, Billerica, MA). Chemicals used for the synthesis of reference materials are described in the supplemental material.

LC-QTOF-MS analysis

LC-QTOF-MS analysis was conducted on an Agilent 1290 Infinity ultra-high-performance LC system coupled to an Agilent 6550 iFunnel QTOF (Agilent Technologies, Kista, Sweden).

A 19-min gradient was used on an Acquity HSS T3 column (150 × 2.1 mm, 1.8 μm, Waters, S) using 0.05% formic acid and 10 mm ammonium formate in water (A) and 0.05% formic acid in acetonitrile (B) as mobile phases at a flow of 500 μL/min. The gradient started with a hold of 1% B for 0.7 min. The metabolites were separated using a linear gradient to 40% B at 13 min followed by a steeper gradient to 95% B at 15 min, which was held until 18 min to wash the column before immediately returning to 1% B until 19 min to re-equilibrate. Injected was 1 μL of sample.

Mass spectrometric data were obtained in positive electrospray ionization mode using Data Dependent Auto MS/MS (gas temperature 150°C, gas flow 18 L/min, nebulizer 50 psig, sheath gas temperature 375°C, sheath gas flow 11 L/min, fragmentor voltage 380 V, collision energy 3 eV at 0 m/z ramped up by 8 eV per 100 m/z, 5 precursors within 200–800 m/z and >5000 counts per cycle, scan rate 6 spectra/s for MS and 10 spectra/s for MSMS, scan range 100–950 m/z in MS and 50–950 m/z in MSMS).

Sample preparation

The 13 urine samples were analyzed by LC-QTOF-MS with and without β-glucuronidase/arylsulfatase treatment to identify potential metabolites.

Sodium acetate buffer (300 μL, 1 m, pH 5) was added to 100 μL urine sample followed by 10 μL β-glucuronidase/arylsulfatase solution in an injection vial. After gentle mixing, samples were incubated for 2 h at 40°C. Similarly, for the non-treated urine samples 100 μL urine was diluted with 310 μL sodium acetate buffer (1 m, pH 5) in an injection vial.

Data analysis

A library of potential metabolite formulae based on mono-, di- and trihydroxylations, carbonylation, carboxylation, methylation, dihydrodiol formation, amide hydrolysis, glucuronidation and N-dealkylation to form norcyclopropylfentanyl, as well as combinations thereof, was compiled.

Samples were searched for metabolites using the library and the Find by formula algorithm of the MassHunter Qualitative Analysis software. Identification was based on presence in urine samples from at least three cases but not in controls, chromatographic peak shape, mass error (less than 5 ppm) and interpretable MS/MS spectra.

Reference material synthesis

Based on preliminary data from urinary analysis and previously published hepatocyte metabolites (7), seven potential metabolites were selected for synthesis: four metabolites monohydroxylated on the phenethyl moiety (2, 3, 4 and β-hydroxy), the catechol 3,4-dihydroxyphenethyl and the two different O-methylation products of the catechol.

To produce the metabolites of interest, a general synthetic route was developed (Figure 1). The amine of 4-piperidone was boc-protected using di-tert-butyl decarbonate and sodium hydroxide in a solvent mixture comprising water:tetrahydrofuran (1:1) as a first step. The resulting carbamate underwent reductive amination by the addition of aniline, acetic acid and sodium triacetoxyborohydride in dichloromethane to form tert-butyl 4-anilinopiperidine-1-carboxylate. In a third step, the free amine was acylated using cyclopropanecarbonyl chloride together with N, N-diisopropylethylamine in dichloromethane to create the normetabolite of cyclopropylfentanyl. This compound was used as a scaffold for the synthesis of seven different potential metabolites of cyclopropylfentanyl. The final steps were N-alkylation using the corresponding bromides with Cs2CO3 in acetonitrile. The bromides were either purchased or synthesized in-house. One synthesis route was reduction followed by an Appel reaction prior to reaction with the normetabolite. Another route was modification after N-alkylation either by reduction using NaBH4 (M11) or by debromination using BBr3 under N2(g) (M5). Synthesis routes can be found in Supplementary Material.

Figure 1

General synthesis route for reference materials. (i) Boc2O, sodium hydroxide, water:tetrahydrofuran (1:1), rt 72 h; (ii) aniline, acetic acid, Na (OAc)3BH, dichloromethane, rt 16 h; (iii) cyclopropanecarbonyl chloride, N, N-diisopropylethylamine, dichloromethane, rt 16 h; (iv) dichloromethane:trifluoroacetic acid (5:1), rt 1 h; (v) corresponding bromide, Cs2CO3, acetonitrile, 60°C 16 h. R1, R2 = H, OH; R3, R4 = H, OH or OH, OMe.

All synthesized potential metabolites including the intermediates were characterized using nuclear magnetic resonance (NMR) spectroscopy and LC-MS. 1H- and 13C-NMR spectra were recorded on a Varian Mercury 300 MHz instrument (25°C in CDCl3). All NMR spectra were calibrated using CDCl3 as reference. MS data were obtained by a Waters SQD 2 Mass Detector using a gradient from 22% to 90% acetonitrile in 10 mm ammonium acetate on an Xbridge C18 column (50 × 4.6 mm, 3.5 μm, Waters). See Supplementary Material for details.

Results

Based on the criteria detailed above, 11 metabolites of cyclopropylfentanyl were identified in the urine samples (Table I and Figure 2). Five out of seven synthesized reference materials had corresponding signals in the urine samples; only 2- and 3-hydroxyphenethyl cyclopropylfentanyl lacked matching metabolite peaks.

Table I

Metabolite areas in hydrolyzed and non-hydrolyzed samples

ID	Average RT a (min)	Metabolite	Formula Mass (exact)	Average mass error (ppm)	Avg Conj b	Peak area (x10 3 ) hydrolyzed (top) and non-hydrolyzed (bottom) samples
						#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	#11	#12	#13
M1	7.81	Norcyclopropylfentanyl	C15H20N2O	1.48	3%	4111	368	348	283	7208	1400	544	410	4895		414	792	26
			244.1576			4118	367	332	268	7212	1366	547	430	4425		366	750	25
M2	8.19	Phenethyl dihydrodiol	C23H30N2O3	−0.74	54%	1642		22	54	1460		37	41	105			201
			382.2256			753			25	705				31			88
M3	8.28	Hydroxyphenethyl glucuronide	C29H36N2O8	−0.77	N/A
			540.2472		Phase II	338			44	1202			24	110			172
M4	8.42	Hydroxy methoxy-phenethyl glucuronide	C30H38N2O9	−0.95	N/A
			570.2577		Phase II	298				595			56				60
M5	9.37	3,4-Dihydroxy-phenethyl	C23H28N2O3	−0.60	92%	1845	76		46	1688		33	133	116			282
			380.21			92	60			121							37
M6	9.43	Phenethyl quinone	C23H26N2O3	1.85	66%	265				237							29
			378.1943			90
M7	9.55	Phenethyl quinone	C23H26N2O3	−0.53	40%	95				141							26
			378.1943			51				92
M8	9.99	4-Hydroxyphenethyl	C23H28N2O2	−1.07	89%	2228	35	114	208	4737	66	115	115	717		52	751
			364.2151			318		33	34	380		34		80			108
M9	10.28	4-Hydroxy-3-methoxy-phenethyl	C24H30N2O3	−0.34	91%	2062		76	41	2525		41	303	83			384
			394.2256			227				160				20			42
M10	10.54	3-Hydroxy-4-methoxy-phenethyl	C24H30N2O3	0.06		63				186			21	30			24
			394.2256
M11	10.72	β-Hydroxy-phenethyl	C23H28N2O2	0.39	36%	139		57	44	372	93	30		425			70
			364.2151			117		50	30	176	50	42				46	48
P	11.72	Cyclopropylfentanyl	C23H28N2O	0.01	3%	4267	114	298	704	3266	420	411	54	2816	74	418	1632	199
			348.2202			4184	107	286	685	3190	412	397	54	2646	79	394	1585	194

Metabolites in bold identified using reference material.

aRT, Retention time

bAvg conj, calculated average conjugation

Figure 2

Metabolic pathway and extracted ion chromatograms of identified metabolites in case #1. All peaks from hydrolyzed sample from case #1 except glucuronide metabolites M3 and M4 that were included from the non-hydrolyzed sample. Metabolites M6, M7, M10 and M11 included as panels with ×10 magnification. Bold metabolites verified by reference materials. *Other potential structures possible.

Except for cyclopropylfentanyl itself and norcyclopropylfentanyl (M1) all metabolites were conjugated in urine based on area differences between the hydrolyzed and non-hydrolyzed samples (metabolites are liberated from conjugates during the hydrolysis). Especially metabolites M5, M8 and M9 were almost entirely conjugated as shown in Table I.

Major metabolites in the hydrolyzed samples were 4-hydroxyphenethyl cyclopropylfentanyl (M8), 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl (M9) phenethyl dihydrodiol cyclopropylfentanyl (M2) and norcyclopropylfentanyl (M1). Norcyclopropylfentanyl was the most abundant metabolite in the non-hydrolyzed samples. Among the minor metabolites, two potential quinone metabolites (M6 and M7) were identified. Cyclopropylfentanyl was found in all samples, both hydrolyzed and non-hydrolyzed.

Discussion

Metabolite identification of metabolites using reference standards

Five metabolites, 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5), 4-hydroxyphenethyl cyclopropylfentanyl (M8), 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl (M9), 3-hydroxy-4-methoxyphenethyl cyclopropylfentanyl (M10) and β-hydroxyphenethyl cyclopropylfentanyl (M11) were identified by reference materials synthesized in-house. The urinary metabolites were matched to the reference materials by accurate mass, retention time and MS/MS spectra match, as shown in Figure 3.

Figure 3

Matching MS/MS spectra of urine samples (top) and the synthesized reference materials (bottom). *m/z 137 and 220 in M8 reference spectra originate from coelution with another reference material.

Metabolite identification of phase I metabolites using MS/MS spectra

Metabolites lacking a matching reference material were identified based on the MS/MS spectra (Figure 4). Norcyclopropylfentanyl (M1) was mainly identified by accurate mass but also based on fragments m/z 69, 84, 162 and 177, indicating the presence of the cyclopropyl ring and the piperidine moiety (accurate masses are given in Figure 4).

Figure 4

Representative MS/MS spectra of cyclopropylfentanyl and metabolites not identified using reference material. GLUC indicate glucuronide. *Other potential structures possible.

Phenethyl dihydrodiol cyclopropylfentanyl (M2) was identified based on the water loss peaks m/z 121 (105 + 16) and 204 (188 + 16) as well as the presence of m/z 69, indicating an unmodified cyclopropyl ring. The exact structure of M2 could not be established; the dihydrodiol can be located in 2,3 or 3,4-position.

Minor metabolites M6 and M7 produced similar MS/MS spectra (Figure 4), consistent with the presence of a quinone structure on the phenethyl moiety that, to our knowledge, has never been reported for any fentanyl analog. M6 and M7 were identified by fragment ions m/z 69 and 228, indicating an unmodified cyclopropyl ring, phenyl and piperidine moiety, as well as m/z 164 indicating a quinone on the phenethyl group. As m/z 132 can potentially be formed from either the phenyl or the phenethyl group, it was of limited value.

Metabolite identification of phase II metabolites

For the glucuronides M3 and M4, MS/MS spectra similar to the corresponding phase I metabolites M8 and M9 were observed. As 4-hydroxyphenethyl cyclopropylfentanyl (M8) is the most abundant monohydroxylated metabolite, it is reasonable to assume that M3 is the 4-hydroxyphenethyl cyclopropylfentanyl glucuronide. Similarly, M4 is likely to be 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl glucuronide as M9 was observed to be the most abundant of the two isomers (M9 and M10).

Metabolites M2 and M5 were also found to be conjugated (Table I). A glucuronidated dihydrodiol metabolite corresponding to M2 was observed at 7.70 min. Also, either of two glucuronidated dihydroxylated metabolites at 8.21 min and 8.49 min could be the conjugate of 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5). MS/MS spectra of these three peaks supported this conclusion, but as they were not detected in three different cases (only in #1 and #5) they were not included.

Cyclopropylfentanyl metabolism

In this study, 11 metabolites of cyclopropylfentanyl were identified in post-mortem urine samples. Two major metabolic pathways were identified: N-dealkylation to form norcyclopropylfentanyl (formed by loss of the phenethyl group) and oxidation of the phenethyl group forming mono- and dihydroxylated metabolites as well as dihydrodiol and quinone metabolites. Oxidized metabolites were further metabolized by methylation and conjugation.

Cyclopropylfentanyl was found in every sample, as shown in Table I. It was the most abundant analyte in 6 out of 13 samples and among the three most abundant analytes in all but one sample (#8). This may indicate that an acute intake of cyclopropylfentanyl resulted in the rapid death of these subjects and does not necessarily reflect the normal metabolism of cyclopropylfentanyl. The most abundant metabolite was norcyclopropylfentanyl (M1), which was detected in 12/13 samples and was the most abundant analyte in 7/13 hydrolyzed samples. Interestingly, in sample #8 norcyclopropylfentanyl was more than seven times larger than the peak of cyclopropylfentanyl, perhaps indicating a repeated intake.

Apart from M1, all metabolites were oxidized on the phenethyl group. Major metabolites included 4-hydroxyphenethyl cyclopropylfentanyl (M8), 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5) and 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl (M9). The 4-hydroxy metabolite was previously reported as a major metabolite of acetylfentanyl and fentanyl after hepatocyte incubation (8). Interestingly, neither 2- nor 3-hydroxyphenethyl cyclopropylfentanyl was observed in any samples, indicating that the hydroxylation is very selective to the 4-position.

Formation of 4-hydroxy-3-methoxy phenethyl cyclopropylfentanyl (M9) from 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5) is potentially catalyzed by the enzyme catechol-O-methyl transferase (COMT) (16). The stereoisomer, 3-hydroxy-4-methoxyphenethyl cyclopropylfentanyl, was also found (M10) but was considerably less abundant. The observed preference of COMT for the 3-position is in line with what has previously been described for the metabolism of other drugs (16).

Other minor metabolites included β-hydroxyphenethyl cyclopropylfentanyl (M11) and phenethyl dihydrodiol cyclopropylfentanyl (M2) as well as two phenethyl quinone metabolites (M6 and M7). Quinone metabolites can be formed from dihydrodiols and/or catechols by a number of different oxidative enzymes (17). Quinone toxicity through binding to proteins and DNA has also been described (17).

Apart from norcyclopropylfentanyl (M1) the metabolites appeared to be 40–91% conjugated in urine based on increased peak area in hydrolyzed samples and further supported by the identification of glucuronide metabolites (M3 and M4). Hydrolysis is recommended for methods aiming to analyze cyclopropylfentanyl metabolites in urine.

Evaluation of the hepatocyte model system

The same LC-QTOF-MS-method was used in this study as in the in the previous metabolism study by Åstrand et al. (7), which made a direct comparison possible. Of seven detected metabolites after hepatocyte incubation, six were also found in this study (M1, M2, M3, M8, M9 and M11). The last metabolite, an N-oxide, was in fact detected in one hydrolyzed and two non-hydrolyzed urine samples, but as this did not represent three unique cases it was not included in this study.

Despite the overall agreement between the results of the hepatocyte study and this study, a few differences were observed: One of the major metabolites in the urine samples, 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5), was only found at low levels after hepatocyte incubation. In addition, β-hydroxy phenethyl cyclopropylfentanyl (M11) was the most abundant monohydroxylated metabolite after hepatocyte incubations but only a minor metabolite in hydrolyzed urine samples.

The results from the urine samples need to be interpreted with caution as no information about the drug consumed was available, and it is possible that impurities present in the ingested drug is an important contributor to observed metabolite levels. To the best of our knowledge, no studies on drug impurities in seized cyclopropylfentanyl samples have been published.

The correlation between urine samples and hepatocytes is in line with other fentanyl analogs (10), but if hepatocytes are used to find analytical targets to analyze urine samples, we recommend that several metabolites to be included and not just the most abundant.

Markers of drug intake

A good urinary marker is abundant, unique to the drug and present in most samples. Based on these criteria and the results of this study, cyclopropylfentanyl in itself appears to be a good urinary marker. Norcyclopropylfentanyl (M1) was the most abundant metabolite but is not unique as it could theoretically be produced from other analogs as well. Alternative markers are 4-hydroxyphenethyl cyclopropylfentanyl (M8), 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5) and 4-hydroxy-3-methoxyphenethyl cyclopropylfentanyl (M9) as they are all abundant, unique and present in a majority of the hydrolyzed urine samples. Synthesis routes for these metabolites were provided in the Supplementary Material.

Evaluation of synthesis

In this study, seven metabolites were synthesized as reference materials allowing for the exact identification of five of the metabolites observed in the urine samples. Furthermore, cyclopropylfentanyl was identified by a commercial reference material, and the exact structure of norcyclopropylfentanyl (M1) could be elucidated from accurate mass and MS/MS spectra. Together, the area of these analytes corresponded to 93% of the total analyte area observed in the hydrolyzed urine samples. This study illustrates how a combination of synthesis of potential metabolites and MS/MS spectra interpretation can be successfully applied to elucidate the exact structure of drug metabolites.

Conclusions

Metabolites of cyclopropylfentanyl were identified in 13 post-mortem urine samples. Cyclopropylfentanyl as parent and the metabolite norcyclopropylfentanyl (M1) were the most abundant analytes. Reference materials were synthesized to identify the exact structure of five metabolites; 3,4-dihydroxyphenethyl cyclopropylfentanyl (M5), 4-hydroxyphenethyl cyclopropylfentanyl (M8), 4-hydroxy-3-methoxy-phenethyl cyclopropylfentanyl (M9), 3-hydroxy-4-methoxyphenethyl cyclopropylfentanyl (M10) and β-hydroxyphenethyl cyclopropylfentanyl (M11). Minor metabolites included two quinone metabolites (M6 and M7), not previously reported for fentanyl analogs.

Abundant metabolites M5, M8 and M9 could, together with cyclopropylfentanyl and norcyclopropylfentanyl, serve as urinary markers of cyclopropylfentanyl intake. To facilitate the manufacture of reference materials, synthesis routes were provided.

Funding

This study was supported by Strategiområdet Forensiska Vetenskaper (Strategic Research Area Forensic Sciences) at Linköping University (S.V. & J.D.) and Eurostar project Psychomics (S.V., J.D., P.K. and H.G.).

Click here for additional data file.