Short-cut route validated for monitoring fentanyl and its metabolite in urine using LC– MS/MS, in a wide concentration range

Background Fentanyl is a highly potent analgesic, used in surgery, frequently abused or used in drug-facilitated crimes (DFC) and in military activities. It is also increasingly used in the treatment of chronic pain (especially in cancer patients). The improper use of transdermal patch forms can cause toxicity and deaths, related to overdose or combined use with other drug substances. Methods are needed for fast, reliable and inexpensive fentanyl detection and we aimed to develop such a method in urine using LC–MS/MS, especially for toxic and fatal concentrations which lack in the literature. Results An LC–MS/MS method has been presented for the co-determination of fentanyl and its main metabolite, norfentanyl in urine. The recoveries of the extraction method were 95(± 6)% and 70(± 9)% for fentanyl and norfenta-nyl, respectively. LOD and LOQ values are 1.7 and 14.0 ng/mL for fentanyl, while they were 20.6 ng/mL and 42.0 ng/ mL for norfentanyl. Conclusion A rapid, sensitive, very practical, inexpensive and a high-recovery analysis method is developed and validated. This is the only fentanyl monitoring LC–MS/MS method in urine having a linearity over a wide range up to 500.0 ng/mL and its success is demonstrated on real samples in the therapeutic drug monitoring of fentanyl and is expected to contribute to clarify intoxications/deaths related to its use.


Background
Fentanyl is a potent, synthetic, narcotic, analgesic opioid with an analgesic activity that is approximately 80 times more effective than morphine and 500 times more potent than meperidine, with fewer side effects [1,2].85% of the fentanyl taken into the body is excreted in the first 72 h; 7% is excreted unchanged (6% urine, 1% feces), 78% is metabolized in the liver (70% urine, 8% feces) [3].Its main metabolite is norfentanyl.
Studies conducted in countries where polydrug use is common, show that a significant group of these individuals who lost their lives due to overdose, combined one or more drugs with the active substance fentanyl [4,5].The transdermal route is superior to both oral and parenteral routes in terms of ease of administration in therapy [6].Dose-related poisoning occurs frequently, as a result of simultaneous use of several transdermal fentanyl bands in home treatment to increase analgesic effect, or in misuse [7].There are many case reports regarding inappropriate use or misuse of fentanyl skin patch content via intravenous, oral, rectal and inhalation [8,9].Defective patches are one of the causes of intoxication [10].Fentanyl has forensic importance as a drug of abuse, which can also be used for suicidal and criminal purposes, besides having toxicological and clinical importance in environmental and occupational exposure, regarding healthcare personnel while administering [11].Environmental exposure might occur via inhalation of powders/aerosols, ingestion, fentanyl patches or items contaminated with fentanyl (which may be present in powder, tablet or liquid forms).Its production and consumption is increasing day by day [12].In recent years, deaths due to fentanyl overdose have been detected among patients hospitalized for chronic pain [6].
In cases of polymorphisms seen in the CYP3A4 gene which metabolizes fentanyl, serious toxicological effects may occur related to fentanyl use [13,14].Therefore, drug monitoring is important for fentanyl.To monitor the levels of fentanyl and its metabolite norfentanyl or to clarify the cases of toxicity/death due to the use of fentanyl transdermal patches in people who use or misuse the patch or in their relatives or health personnel who are exposed to them; a simple and rapid method in urine, with low detection limit, higher recovery, having a wide linearity range, was developed using LC-MS/ MS.Although the extraction method was modified from study of Coopman et al. [15], differently here, steps were reduced using undiluted urine; recovery for norfentanyl was increased twice; linearity range was increased up to a very high amount as 500 ng/mL; and a 2.5 times enrichment was performed, while a lower injection volume was used.Papaverine hydrochloride was chosen as the internal standard (IS) and a 2.5-min ultra-fast LC-MS/ MS analysis was performed using C 18 poroshell column.The whole validated method is for quantitative purposes with given LOD and LOQ values.This method is the first short-cut fentanyl monitoring method in the literature with the widest concentration range, requiring no dilution at toxic concentrations.

Sample collection
Urine samples were collected from five healthy individuals (between ages 24 and 36 years old) with no drug use and with no known chronic disease, and mixed in a pot to form a blank urine pool for method development and validation.To test the method in real cases, approximately 50 mL of urine samples was obtained from three inpatients (aged between 54 and 66), two of whom were female, in cooperation with Istanbul University-Cerrahpasa Medical Faculty, Department of Algology, for analysis.Two of the participants had diagnosed with servix cancer (body weight: 42 and 65 kg) and one of them with soft tissue cancer (body weight: 73 kg) in the last 2 years and all were treated with transdermal fentanyl patches to overcome violent pain.None had tobacco or alcohol use and all were unrelated from the point of kinship and did not use any other medication.The daily doses of the amount of fentanyl used by patients varied between 50 and 100 µg/h.All collected urine samples were placed in sterile, plastic urine containers and were stored at + 4 °C until analysis.Standard solutions were spiked to these urine samples, extracted and analyzed.

Preparation of the standard solutions and urine samples
Analyte and IS solutions were prepared using methanol at final concentrations of 500.0,5000.0,12,500.0,50,000.0ng/mL and added freshly to blind urine samples in predetermined volumes, not exceeding 20.0 µL.A 20.0 µL IS solution of 20.0 µg/mL was also added.Finally, methanol in varying amounts was added to each sample in order to standardize the spiking volume as 40.0 µL.Blind urine samples were also studied.All stock and test solutions were prepared freshly.

Optimization of the extraction method
The extraction method was modified from the liquid-liquid extraction (LLE) method of Coopman et al. [15] and optimized.In the modification and optimization studies; the effect of mixing time, sample volume and the number of extraction steps on the recovery efficiency were investigated.Unlike the former study, undiluted urine was used for extraction.The effect of salt on extraction efficiency was investigated through testing NaF, Na 2 SO 4 and K 2 CO 3 in the extraction with 7.00 mL of n-hexane/ ethylacetate (7:3, v:v).When NaF and Na 2 SO 4 were used, the phase separation was not clear, and K 2 CO 3 provided the best phase separation.Since fentanyl and norfentanyl were weak bases, the nonionized fractions of the analytes and the recovery were increased after the added urine was alkalized (to pH = 11.50-12.00)with K 2 CO 3 solution.The effect of mixing time on the recovery of fentanyl in LLE was also investigated; trying 3, 5 and 7 min extraction times.7 min was observed to give much higher recovery than the Coopman's method.Sample volumes of 1.00 and 2.00 mL were tried and the best recovery was obtained with 1.00 mL sample volume.Also a twostep extraction method was tried with a total of 7.00 mL The optimized sample preparation method below had higher precision and higher analytically acceptable recoveries than the unmodified form: 1.00 mL of urine was placed in a polypropylene test tube, 0.363 M K 2 CO 3 was added till a pH between 11.5 and 12.0 was obtained.After 30 s vortexing, 7.00 mL of n-hexane/ethylacetate (7:3, v:v) mixture was added and vortexed for 7 min.After 5 min centrifugation at 5000 rpm, the upper phase was evaporated to dryness at 35 °C under nitrogen, reconstituted in 400.0 μL methanol and transferred to vials for analysis in LC-MS/MS.
All measurements were performed, wherever possible, at low temperatures (autosampler: at 4 °C, evaporation: at 35 °C) and in a short period of time.

Validation
The chromatograms of solvent, extracted blank and spiked samples (n ≥ 6) were compared to determine the specificity of the method.It was checked whether there was any interference from the solvent and the matrix, at the ion channels and retention times of the peaks of fentanyl and norfentanyl.In addition, in order to provide the selectivity, characteristic MRM ion channels specific to each of the analytes and IS were created.Each standard was injected one by one to monitor whether their peaks interfere with each other's ion channels to demonstrate the selectivity.Data evaluation and calculations of validation parameters were performed using Microsoft Excel.
A matrix-matched calibration technique was used for the compensation of matrix effects: Spiked matrix samples were prepared in increasing concentrations between 1.0 and 1250.0 ng/mL for construction of the calibration curves.Least squares method was used to create the calibration curves.Standard deviations obtained from at least two injections of each of three samples (n ≥ 6) for each concentration level were calculated.The relative standard deviation (RSD%) values for each calibration level were used in LOQ estimation via Eurachem method [16], which is the most strict method utilizing the reasonable and a gradual decrease of the repeatabilities in the y-axis of the Eurachem graph versus increasing concentration and regarding 15% RSD as the LOQ.This was estimated through plotting RSD% values versus concentrations of the spiked samples analyzed during the matrix-matched calibration.LOD was calculated using Eq. 1 [17]: s: standard deviations of the analyte area/IS area ratios.
Recovery studies were performed at low, medium and high concentrations of the linear range of fentanyl and norfentanyl for urine samples, each with at least six replicates.Average recovery rates along with their precision values were calculated from Eq. 2 [18]: The matrix effect was evaluated by comparing spiked matrices with the standard solutions in methanol (n ≥ 6) and was studied for full concentration scale in the calibration graph (Eq. 3) [19]: (3) Matrix Effect, % = slope analytical curve standard in matrix − slope analytical curve standard in solvent slope analytical curve standard in solvent ×100

Results
In this study, a fast and simple LC-MS/MS method for fentanyl and its metabolite norfentanyl in urine is developed and validated.The sample preparation method was modified from Coopman et al. [15] and optimized.In the method developed for screening and confirmation, the analytes were determined in 1.00 mL urine sample, with a very fast (2.5 min) isocratic elution in LC-MS/MS, following a 30-35 min LLE.LLE was performed at a pH between 11.5 and 12.0 (K 2 CO 3 ) using 7.00 mL of n-hexane/ethylacetate solution, in 7 min and the organic phase was enriched 2.5 times for analysis in LC-MS/MS.
The retention times for fentanyl and norfentanyl were 1.73(± 0.00) and 1.38(± 0.01) minutes.MRM transitions were 337.0 > 188.0 m/z for fentanyl and 233.2 > 150.1 m/z for norfentanyl.The retention time for IS was 1.73(± 0.00) minutes, while its MRM transition was 340.1 > 202.0 m/z.The validation was performed by means of specificity, selectivity, linearity, linear range, recovery, LOD, LOQ and matrix effect.

Validation of the optimized method
In the validation of the optimized method, the selectivity was confirmed by the determination of the analytes in their characteristic ion channels.As a result of the comparison of the chromatograms of the analytes in urine extracts, with the methanol and blank urine chromatograms; it was observed that no interference from the solvent and the blank urine existed in the retention times of the analytes, so the specificity was also confirmed (Fig. 1).
Calibration of the whole method, including sample preparation and instrumental method, was performed with matrix-matched calibration technique at five different concentrations.The equations of the calibration curves, linear ranges, regression limit value, LOD and LOQ values of the instrumental method regarding standard fentanyl and norfentanyl in solutions and in urine matrix are shown in Table 2. Calibration curves of the standard solutions as well as the matrix-matched calibration curves for fentanyl and norfentanyl, along with their r 2 values are demonstrated in Fig. 2. Matrix effect calculated for each analyte (n ≥ 6) was found as 6.25% for fentanyl showing an ion enhancement and -20% for norfentanyl indicating anion suppression.The validation results obtained in standard solutions and sample extracts fit for the purpose for analysis in tablet and injectable solutions and drug monitoring/toxicological analysis in urine, respectively.
An average recovery of 95(± 6) % was obtained for fentanyl and 70(± 9) % for norfentanyl, regarding low, mid and high concentrations (Table 3).Recovery findings were found to be within acceptable limits.

Real examples
The method was applied to urine samples of three patients treated with fentanyl in Istanbul University-Cerrahpasa Medical Faculty, Department of Algology.The analysis results obtained using LC-MS/MS are shown in Table 4. Since the amount of substance in the urine of the patients is variable, blank methanol was analyzed between samples, in order to prevent contamination.The chromatograms obtained after each blank methanol analysis were examined and it was seen that there was no carry over between the analyses.Fentanyl and norfentanyl concentrations were calculated using calibration graphs.The lowest amount of fentanyl detected was 32.6 ng/mL and the highest amount was 111.3 ng/mL.

Discussion
In the literature, deaths due to fentanyl absorption in high doses have been detected in patients treated with fentanyl transdermal patches for chronic pain [6].Various LC-MS/MS methods have been developed for the determination of fentanyl and norfentanyl, in urine and other biological materials, as well as in formulations [15,20].Although our basic extraction method was modified from Coopman's work [15], the recovery obtained here is almost doubled for norfentanyl (raised from 40 to 70%).Wider linearity range up to 500 ng/mL was provided (so that no dilution will be required in higher concentrations), which does not exist in any of the studies encountered.Since it is important to see higher concentrations in serious toxicities, wider linearity range will provide immediate results without dilution steps and without any change of matrix effect.Also, in our method, urine was used without dilution (number of steps were decreased),   mixing time, sample volume and elution and drying temperatures were optimized, papaverine hydrochloride was chosen as the IS, and a 2.5 times enrichment was performed, while a lower injection volume was used.In this study, a quantitative method was validated by calculating LOQ values in standard solutions and in urine.The effect of different salts was also examined.Furthermore, this method is very fast compared to the literature methods, especially when the 2.5 min chromatographic elution is considered.An average of 95(± 6)% recovery was obtained for fentanyl and all RSD% values were < 15.0%.Also, our modified extraction method is simpler, faster [21-23] and 2.5 min LC-MS/MS method is shorter than all the equivalent literature methods regarding fentanyl   and norfentanyl, except the method of Mahlke et al., where fentanyl was eluted in 1.68 min (1.73 in our study) and norfentanyl was eluted in 1.56 min (1.38 in our study).However, the linear range in most of the studies including Mahlke et al. were in maximum up to 10.0 or 50.0 ng/mL, while our method can detect higher and toxic concentrations without any dilution [10,[23][24][25][26][27].
Verplaetse and Tytgat [17] have used solid phase extraction (SPE) for determination of fentanyl and norfentanyl in urine and have given a linearity up to 5 ng/ mL and 10 ng/mL, respectively [11].They also developed another SPE method with excellent recoveries for nine narcotic analgesics and metabolites including fentanyl and norfentanyl, with a 7.2 min elution time for fentanyl and 10 and 40 ng/mL as the highest levels of given linearity range, for the drug and its metabolite [12].Cunha et al. [28] have performed a microextraction by packed sorbent, with a chromatographic run of 10.3 min.Recovery was at least 27% for all analytes and 13% for norfentanyl.Eckart et al. [29] have developed an automated SPE method for a wide range of opioids with a highest calibration level 10 ng/mL, where the recovery for fentanyl and norfentanyl was between 62 and 78% and the elution times were 21.3 min and 10 min.As a result, the calibration range of our study is wider than the published equivalent studies, and our method is faster, easier, besides presenting high recoveries.
Fentanyl test strips are also used for testing the presence of fentanyls in illicit drug products.However, most fentanyl test strips are based on competitive lateral flow immunoassays on simple paper-based devices [30] and there is a considerable lack of knowledge about the selectivity and sensitivity of testing with urine fentanyl strips [31].Since the test results are highly concentrationdependent, detection of lower concentrations can be problematic [30].The fentanyl test strips do not measure the exact fentanyl concentration and do not provide results that differentiate between or among fentanyl and any of the analogs present in a sample [32].Most of the strips give only qualitative results, but with a validated fast chromatographic method, exact concentration can be determined quantitatively and sensitively.An additive effect would occur in case of presence of acetylfentanyl, acrylfentanyl, furanylfentanyl and butyrfentanyl at low concentrations of fentanyl (10 ng/mL), thus false positive results may be obtained.It was only demonstrated on the Nal van Minden test, but this additive effect is expected to be a general principle for all test strip brands (false positive).Four brands of fentanyl test strips (Rapid Response, One Step, Nal van Minden and Rapid Self Test) were examined using single-component drug solutions [30].21-24 of the 28 fentanyl analogues were tested.The effect of co-presence of heroin or ascorbic acid on test results was also examined and one of the test strip brands gave false positive results in the presence of ascorbic acid.With strip, fentanyl and norfentanyl cannot be distinguished from other fentanyl analogues, but a chromatographic analysis can selectively determine fentanyl.Since a clean-up step is applied in this method, the selectivity is much more higher than testing with fentanyl strips.
Different amounts of fentanyl and norfentanyl found with this method in the urine samples of patients who were treated with patches containing the same dose, arouse the conviction that individual characteristics (gender, age, enzyme polymorphism, foodstuffs, other drugs taken together with fentanyl, urine concentration, etc.) or periodical/non-periodical applications might be the reasons.Besides these inter-individual differences, from time to time, differences within individuals are reported during the treatment [33].The patient with the highest value has received the last fentanyl dose (100 μg/hour) two days before the urine sample was taken, and the drug administration was continued at 72-h intervals.The lowest amount of norfentanyl detected was 151.3 ng/mL and the highest amount was 267.2 ng/mL.According to the information of the patient with the lowest level of norfentanyl; a dose of 50 μg/hour was administered, and it was not repeated at 72-h intervals due to the patient's disrupting the medication.

Conclusion
Considering the risk of intoxication/death that may be encountered during treatment or because of misuse of fentanyl, a fast (30-35 min sample preparation + 5 min elution time in total), validated, sensitive LC-MS/MS analysis method with high recovery is presented here for the simultaneous determination of fentanyl and its metabolite in urine; to benefit in therapeutic drug monitoring, emergency toxicology and in forensic cases as toxicity/death caused by this drug and in supporting the presence of polymorphism of CYP3A4 gene (metabolizing fentanyl).The recovery obtained by the modified extraction method is almost doubled for norfentanyl and the upper linearity range was increased to at least10 times that of the former methods.Since this method has the widest linearity range in the literature, it is employable in a wide concentration range and no dilution will be required in higher concentrations.Especially, overdilution of the matrices will change the matrix intensity and errors may arise in the results.We suggest that this analysis method should be tested in future studies for the determination of new and more potent fentanyl derivatives in the market.Since it is important to determine higher concentrations in serious toxicities, wider linearity

Fig. 1
Fig. 1 MRM Chromatograms of a fentanyl and b norfentanyl, added to the blind urine matrix before extraction at concentrations of 250.0 ng/mL and c Papaverine (IS) at 5.0 ng/mL

Fig. 2
Fig. 2 Calibration curves of fentanyl (a), norfentanyl in neat solutions (b) and matrix-matched calibration curves for fentanyl (c) and norfentanyl (d) in urine

Table 1
MRM table of the instrument

Table 2
Linearity, linear ranges, calibration equations, LOD and LOQ values for fentanyl and norfentanyl regarding the whole method r 2 > 0.99; a ng/mL, number of analyses n ≥ 6, RSD% < 15

Table 3
Recovery and precision values for fentanyl and norfentanyl regarding the whole method at low, mid and high concentrations

Table 4
Concentrations of fentanyl and its metabolite found in the urine of the patients after the given doses (n ≥ 3) a RSD % for fentanyl ≤ 6 b RSD % for norfentanyl ≤ 12