Chiral Liquid Chromatography-Tandem Mass Spectrometry Analysis of Superwarfarin Rodenticide Stereoisomers – Bromadiolone, Difenacoum and Brodifacoum – in Human Plasma

Abstract

Superwarfarins are second-generation long-acting anticoagulant rodenticides that can cause unintended human and wildlife toxicity due, in part, to their prolonged half-lives. Commercially available superwarfarin rodenticides are synthesized as racemates with two asymmetric carbons, producing four stereoisomers. To support studies of human plasma half-lives of individual superwarfarin stereoisomers, a method was developed based on LC-MS/MS to separate and quantify stereoisomers of the commercially important superwarfarins bromadiolone, difenacoum and brodifacoum. Human plasma samples were prepared using protein precipitation and centrifugation. Chiral-phase HPLC separation was carried out on-line with tandem mass spectrometric quantitative analysis of the eluting stereoisomers using selected-reaction monitoring with positive ion electrospray on a triple quadrupole mass spectrometer. All four stereoisomers of each superwarfarin were resolved within 12.5 min with calibration curves spanning 2 – 3 orders of magnitude and lower limits of quantitation between 0.87 – 2.55 ng/mL. This method was used to determine the half-lives of superwarfarin stereoisomers in plasma from patients who had inhaled synthetic cannabinoid products contaminated with superwarfarins. These data may be used to guide the development of safer next generation anticoagulant rodenticides stereoisomers.

1. Introduction

Based on the discovery of the anticoagulant dicoumarol in 1939, the derivative warfarin was commercialized as an anticoagulant rodenticide in 1948 [1,2]. The overuse of warfarin resulted in the development of resistance among rodents by the 1960s [3]. More potent second generation derivatives bromadiolone (BDL), difenacoum (DFC), flocoumafen (FCF), brodifacoum (BDF) and difethialone (DFT) (Figure 1), also known as superwarfarins, were synthesized in the 1970s and 1980s and shown to be effective against warfarin-resistant rodents [4–6]. Dicoumarol, warfarin and superwarfarins share a 4-hydroxy coumarin scaffold and function by inhibiting the vitamin K 2,3-epoxide reductase (VKORC1), disrupting the vitamin K cycle and preventing the activation of clotting factors [7–9].

Figure 1.

Chemical structures of bromadiolone (BDL), difenacoum (DFC), brodifacoum (BDF), difethialone (DFT) and flocoumafen (FCF). Each superwarfarin contains two chiral carbons forming 4 isomers (2 pairs of diastereomers).

Superwarfarins have a superior single dose effectiveness and 100-fold greater potency compared to warfarin; characteristics which have contributed to poisonings of humans and non-target wildlife [10–13]. Approximately 10,000 human cases of superwarfarin poisonings are reported in the United States each year, with the majority occurring in children [14–16]. During the last decade, there have been multiple reports of intentional human poisonings by superwarfarins [14]. In 2018, there was an outbreak of poisonings in the Midwest of the United States in which over 300 patients were hospitalized due to inhalation of synthetic cannabinoids contaminated with three different superwarfarins [17,18]. Treatments to restore blood coagulation following superwarfarin poisoning include infusion of fresh frozen plasma, red cell transfusion, intravenous prothrombin protein complex, intravenous vitamin K1, and oral vitamin K₁ [12]. Due to enterohepatic recirculation and slow metabolism, the half-lives of superwarfarins are long, accumulation can occur following multiple small doses, and multiple organ damage can occur including renal injury [12,19,20].

In an effort to mitigate human exposure to superwarfarins, the following procedures are being pursued: treatments which accelerate superwarfarin elimination from the body [21], decreasing superwarfarin rodenticide concentration [22], replacing superwarfarin with cholecalciferol (vitamin D₃) [23], combining two rodenticide active ingredients in a single product [24–26], and investigating superwarfarin stereoisomers [27–29]. In view of the above potential deleterious actions of superwarfarin [12], an effective superwarfarin rodenticide is needed with curtailed toxic accumulation in humans and non-target wildlife [27,30].

Superwarfarins possess two chiral carbons and are synthesized as racemates, containing four stereoisomers, which can selectively interact with VKORC1 and other proteins, such as drug metabolizing enzymes [31,32]. For example, enantiomers of warfarin and phenprocoumon are both effective anticoagulants, but the (S)- enantiomer of each is 3 – 5 times more potent than the corresponding (R)- enantiomer, even though the more potent (S)- enantiomers are cleared more quickly [33–35]. In addition, significant differences in half-lives have been reported for pairs of cis- and trans- diastereomers of superwarfarins (Figure 1) in rats, rabbits and humans [18,27,29,36–41]. Recently, Lefebvre et. al. determined that the half-lives of the trans-DFT enantiomers were 6.0 and 69.3 hours, respectively, while the half-lives of the cis-DFT enantiomers were 25.4 and 82.3 hours, respectively [30]. Limited studies have been performed concerning the relative toxicity of superwarfarin stereoisomers.

To explore the half-lives of specific superwarfarin stereoisomers further, we developed a high pressure liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) based method to separate and measure all four stereoisomers of the most commonly used rodenticides, BDL, DFC and BDF, using chiral chromatography in a single analysis. Enantiomers possess nearly identical chemical properties, with the exception of their interaction with plain polarized light, and cannot be separated using an achiral stationary phase. Several HPLC columns with enantiomerically pure stationary phases were screened from Phenomenex and AZYP USA in order to obtain a chiral column with the selectivity required to separate the four stereoisomers of BDL, DFC and BDF. This method was applied to plasma samples from patients who inhaled synthetic cannabinoids tainted with superwarfarins during the Illinois outbreak in early 2018 [17,41,42].

2. Experimental

2.1. Chemicals and reagents

LC-MS grade acetonitrile, methanol and formic acid as well as the superwarfarins BDL, DFC, and BDF were purchased from Sigma-Aldrich (St. Louis, MO, USA). The internal standard BDF-d₄ was purchased from Toronto Research (Ontario, Canada). The concentrations of each stereoisomer (Figure 1) were determined by obtaining their ratios from calibration standards and multiplying the ratios by the total concentration of each superwarfarin (Table 1). Blank pooled human plasma was purchased from Lampire Biological Laboratories (Pipersville, PA, USA). Plasma samples from subjects were obtained from the OSF Saint Frances Medical Center (Peoria, IL, USA). Purified water was prepared using a Milli-Q water purification system (Millipore, MA, USA).

Table 1.

UHPLC retention times, MS/MS collision energies (CE) and precursor/product ion selected reaction monitoring (SRM) transitions.

Superwarfarin	Retention time (min)	Precursor ion	Quantifier ion		Qualifier ion
		m/z	m/z	CE (V)	m/z	CE (V)
cis-1 BDL	2.9	511	251	−23	277	−21
cis-2 BDL	3.1
trans-1 BDL	3.4
trans-2 BDL	4.0
cis-1 DFC	3.9	445	179	−31	257	−20
cis-2 DFC	4.3
trans-1 DFC	6.1
trans-2 DFC	7.7
cis-1 BDF	4.4	525	337	−23	178	−35
cis-2 BDF	4.7
trans-1 BDF	6.4
trans-2 BDF	7.8
cis-1 BDF-d4	4.3	529	337	−23	178	−41
cis-2 BDF-d4	4.6
trans-1 BDF-d4	6.3
trans-2 BDF-d4	7.7

2.2. Preparation of human plasma

Human plasma, which had been collected in vacutainers and stored at −80 °C, were thawed, vortexed and centrifuged at 20,000 × g for 20 min at 4 °C [39]. Protein precipitation of the plasma was carried out by adding 400 μL of 10% methanol in acetonitrile containing BDF-d₄ (37.5 ng/mL) to 100 μL of plasma. After vortex mixing for 5 min and centrifugation at 20,000 × g for 15 min at 4 °C, the supernatant (400 μL) was removed and evaporated to dryness under vacuum. The residue was reconstituted with 50 μL of 50% (v/v) aqueous methanol, vortexed for 5 min and centrifugated at 20,000 × g for 15 min at 4 °C immediately before analysis using HPLC-MS/MS.

2.3. Preparation of calibration standards and quality control samples

Stock solutions of BDL, DFC, BDF and BDF-d₄ were prepared in DMSO (1 mg/mL) and stored at −20 °C. Working solutions ranging from 40 ng/mL to 40,960 ng/mL for BDL, DFC and BDF were prepared from stock solutions using serial dilution in 50% (v/v) aqueous acetonitrile and stored at −20 °C. Calibration standards were prepared by spiking 95 μL of blank plasma with 5 μL of working solution followed by extraction, as stated above. Quality control (QC) samples were prepared identically to the standards at the lower limit of quantitation (LLOQ), low, medium and high concentrations. The LLOQ was defined as a signal-to-noise (S/N) ratio ≥ 10, the low concentration QC was three times the LLOQ, the medium concentration QC was 40% of the upper limit of quantitation (ULOQ), and the high concentration QC was 80% of the ULOQ for each analyte. The lower limit of detection (LLOD) was defined as a S/N ratio of 3.

2.4. Chiral column selection

Selectivity of functionalized cellulose and amylose-based chiral columns from Phenomenex (Torrance, CA) were screened using the vendor recommended isocratic elution of water and acetonitrile (20:80, %v/v) both containing 0.1% acetic acid. The columns included the Lux Cellulose-1 (cellulose tris(3,5-dimethylphenylcarbamate)), Lux Cellulose-2 (cellulose tris(3-chloro-4-methylphenylcarbamate)), Lux Cellulose-3 (cellulose tris(4-methylbenzoate)), Lux Cellulose-4 (cellulose tris(4-chloro-3-methylphenylcarbamate)), Lux iCellulose-5 (cellulose tris(3,5-dichlorophenylcarbamate)), Lux Amylose-1 (amylose tris(3,5-dimethylphenylcarbamate)), and Lux iAmylose-1 (amylose tris(3,5-dimethylphenylcarbamate)), each with dimensions of 100 × 4.6 mm, 5 μm.

Chiral columns from AZYP (Arlington, TX) were also evaluated for their chromatographic selectivity, which included the Teico Shell (teicoplanin), Vanco Shell (vancomycin), Nico Shell (modified vancomycin), Q-Shell (quinine), CD Shell-RSP (hydroxylpropyl β-cyclodextrin), Larihc Shell-P (Isopropyl cyclofructan 6), Larihc CF6-RN (R-naphthylethyl cyclofructan 6) and Larihc CF7-DMP (3,5-dimethylphenyl cyclofructan 7), each with dimensions of 100 × 4.6 mm, 2.7 μm. The Teico Shell, Vanco Shell, Nico Shell and Q-Shell chiral columns were preliminarily screened using the vendor recommended isocratic elution of methanol with 0.1% formic acid. The CD Shell-RSP, Larihc Shell-P, Larihc CF6-RN and Larihc CF7-DMP chiral columns were screened using acetonitrile, methanol and formic acid (60:40:0.2, %v/v).

2.5. Chromatography

Separations were carried out using a Shimadzu (Kyoto, Japan) Nexera UHPLC system equipped with an AZYP (Arlington, TX, USA) Q-Shell (100 × 2.1 mm, 2.7 μm) chiral column. Both mobile phases contained water/methanol (5:95; v/v) with mobile phase A containing 0.1% formic acid and mobile phase B containing 0.5% formic acid. A linear gradient from mobile phase A to B was used at a flow rate of 0.4 mL/min as follows: 0.0 – 4.0 min, 0% B; 4.0 – 7.5 min, 0 – 100% B; 7.5 – 10.0 min, 100% B; 10.0 – 10.1 min, 100 – 0% B; 10.1 – 12.5 min, 0% B. The column temperature was 40°C, and the injection volume was 10 μL.

2.6. Tandem mass spectrometry

Superwarfarins eluting from the chiral column were measured using a Shimadzu LCMS-8060 triple quadrupole mass spectrometer operated at unit resolution. Nitrogen was used for nebulization (2 L/min), as the heating gas (10 L/min) and as the drying gas (10 L/min). The temperatures of the interface, the desolvation line and the heat block were 300, 250 and 400 °C, respectively. Superwarfarins were measured using positive ion electrospray, collision-induced dissociation with argon gas at 230 kPa, and selected reaction monitoring (SRM). The SRM transitions for MS/MS quantification (quantifier) and for quality control (qualifier) for each analyte are shown in Table 1.

2.7. Calibration curve linearity

Calibration curves were constructed in duplicate using 7 calibration standards for cis-BDL, 6 for trans-BDL and 9 for DFC and BDF. The natural log of the peak area ratio of analyte to internal standard was plotted against the natural log of the analyte concentration. Calibration curve linearity was determined using least squares linear regression analysis. No weighing factor was applied. Data analysis was carried out using Shimadzu Lab Solutions software (Ver. 5.97).

2.8. Selectivity and matrix effect

Six lots of blank human plasma were analyzed to evaluate the method selectivity, defined as the extent to which sample components interfered with the SRM signal of each superwarfarin. Significant interference was defined as the detection of any peak at the retention time of any superwarfarin stereoisomer with an area greater than 20% of the analyte at the LLOQ or greater than 5% of the internal standard. Matrix effects were assessed by comparing the peak areas of spiked blank matrix to the peak areas of spiked neat injection solvent at low and high analyte concentrations, using six lots of blank plasma. Deviation in peak area of >15% compared to neat injection solvent would be considered significant.

2.9. Precision and accuracy

Repeatability (intra-day variability), intermediate precision (inter-day variability) and accuracy were evaluated by performing 3 separate analyses on 3 consecutive days. Each analysis included a calibration curve with 5 quality control (QC) samples at the LLOQ, low, medium, and high concentrations. Repeatability and intermediate precision for each concentration were defined as the relative standard deviation (RSD) of replicate QCs. Accuracy was defined as the percent difference between the mean of replicate QCs and the true nominal value at each concentration.

2.10. Recovery

The recoveries of each superwarfarin from human plasma were as described previously [39]. Briefly, the recoveries of superwarfarin from human plasma after protein precipitation were determined at low, medium and high concentrations by comparing the peak area ratios of experimental to theoretical samples. Experimental samples consisted of standards spiked into blank plasma, processed and reconstituted with neat injection solvent containing internal standard. Theoretical samples consisted of neat solvent spiked into blank plasma, processed and reconstituted with standards in the injection solvent containing internal standard.

2.11. Stability

Stabilities of BDL, DFC, FCF, BDF, and DFT have been reported previously [39]. Briefly, short-term stabilities were assessed over 24-h at low and high concentrations at room temperature on the benchtop and at 4 °C in the autosampler. Long-term stabilities at low and high concentrations were assessed over 30-days during storage at −20 °C and −80 °C.

2.12. Data analysis

For each stereoisomer, the plasma superwarfarin concentration over time was plotted using Phoenix 64 WinNonlin Modeling Software (ver. 8.2). Best fit linear regression was performed on the set of time points and the slope was used to calculate the half-lives for each superwarfarin stereoisomer. Statistical analysis was performed with GraphPad Prism software (ver. 7.04). Outliers were assessed using the Grubb’s test (α = 0.05), and the comparison of mean stereoisomer half-lives was conducted using the unpaired Student T-test (α = 0.05).

3. Results

3.1. LC-MS/MS method development

Several chiral columns from Phenomenex (Lux Cellulose-1, Lux Cellulose-2, Lux Cellulose-3, Lux Cellulose-4, Lux Amylose-1 and Lux iAmylose-1) and AZYP USA (Teico Shell, Vanco Shell, Nico Shell, Q-Shell, CD Shell-RSP, Larihc Shell-P, Larihc CF6-RN and Larihc CF7-DMP) were evaluated for the separation of superwarfarin enantiomers. The Phenomenex Lux Cellulose-2, Lux Cellulose-3, Lux Cellulose-4, Lux iCellulose-5, Lux Amylose-1, Lux iAmylose-1 and the AZYP Teico Shell, Vanco Shell, Nico Shell, CD Shell-RSP, Larihc Shell-P, Larihc CF6-RN, and Larihc CF7-DMP columns did not separate the superwarfarin stereoisomers, which eluted from the columns as a single peak or a broad peak with shoulders. The Phenomenex Lux iCellulose-5 showed adequate selectivity to separate at least partially all four stereoisomers of each superwarfarin. Despite attempts to optimize mobile phase conditions for the Lux iCellulose-5 column, baseline resolution of the four enantiomers of BDF, DFC and BDL could not be achieved within 60 min.

Baseline resolution of superwarfarin enantiomers was achieved in 40 min using the AZYP Q-Shell chiral column with the vendor recommended isocratic mobile phase conditions. This initial isocratic elution was optimized using various mobile phases, including water, methanol and acetonitrile, and additives of formic acid or acetic acid. Fast elution was obtained using acetonitrile, but chromatographic resolution was low. Longer retention times and increased chromatographic resolution were observed when using aqueous methanol. The addition of formic acid drastically decreased peak broadening while a loss of chromatographic resolution was observed when adding acetic acid. As with all isocratic chromatography, band broadening was observed for late eluting peaks. To minimize this band broadening, a linear gradient of increasing formic acid was used from 0.1% to 0.5% in methanol/water (95:5, v/v). This gradient also shortened chromatographic separation time from 35 to 12.5 min, including column equilibration (Figure 2).

Figure 2.

Positive ion electrospray chiral HPLC-MS/MS SRM chromatograms of BDL (100 ng/mL), DFC (400 ng/mL), BDF (800 ng/mL) and BDF-d₄ (internal standard, 150 ng/mL) standards spiked into blank human plasma. The stereoisomeric ratio of each superwarfarin standard is as follows: (cis-1:cis-2:trans-1:trans-2): BDL (36:39:13:12), DFC (22:23:29:26), BDF (17:21:32:30) and BDF-d₄ (30:36:17:17).

Protein precipitation using a mixture of acetonitrile and methanol was used for sample preparation instead of liquid-liquid extraction with chloroform and acetone as described previously [43]. This approach prevented co-extraction of plasma lipids, which could accumulate on the column and necessitate an extra column washing step.

Optimization of tandem mass spectrometry parameters including selection of precursor ion mass, product ion mass and collision energy was carried out for each superwarfarin (Table 1). Positive ion electrospray produced abundant protonated molecules of DFC and BDF, which were used as precursor ions. Due to abundant in-source loss of water for BDL, the [MH-H₂O]⁺ ion was used as the precursor ion during SRM. For BDL and BDF (including the internal standard BDF-d₄), the heavier bromine isotope of each precursor ion was used, because it produced greater signal-to-noise than did the lighter bromine isotope.

The elution order of superwarfarin stereoisomers (cis-1, cis-2, trans-1 and trans-2; Figure 1) were determined by comparing the stereoisomeric ratios to the diastereomeric ratios that had been obtained by using achiral separations [29,36,39,43,44]. Diastereomers are not mirror images of one other and have different physicochemical properties, which facilitates their separation using achiral reversed phase chromatography. The cis:trans diastereomeric ratios for BDL, DFC, BDF and BDF-d₄ have been reported to be between 70:30–90:10, 45:55–80:20, 40:60–65:35 and 65:35–75:25, respectively [29,36,39,43,44]. The stereoisomeric ratios (cis-1:cis-2:trans-1:trans-2) of BDL, DFC, BDF and BDF-d₄ standards were determined to be (36:39:13:12), (22:23:29:26), (17:21:32:30) and (30:36:17:17), respectively (Figure 2). When combined, these stereoisomeric ratios correspond to diastereomeric ratios of 75:25, 45:55, 38:62, and 66:44, respectively. Although modification of chromatographic conditions can change the elution order of analytes, comparison of these measured but simplified stereroisomeric ratios to the reported superwarfarin diastereomeric ranges is consistent with an elution order of cis-enantiomers followed by trans-enantiomers [45–47].

3.2. Linearity and lower limit of quantitation (LLOQ)

Calibration curves for BDL, DFC and BDF stereoisomers were plotted using log transformation of both analyte concentration and peak area ratios [48]. The curves were linear across 2 – 3 orders of magnitude with coefficients of determination (R²) of >0.999 for each superwarfarin stereoisomer (Table 2). Experimentally determined lower limits of quantitation (LLOQ) were between 0.87 and 2.55 ng/mL with upper limits of quantitation (ULOQ) between 30.40 and 651.98 ng/mL and lower limits of detection (LLOD) between 0.43 and 1.27 ng/mL. Experimentally determined lower limits of quantitation (LLOQ) were between 0.87 and 2.55 ng/mL with lower limits of detection (LLOD) between 0.43 and 1.27 ng/mL and upper limits of quantitation (ULOQ) between 30.40 and 651.98 ng/mL.

Table 2.

Coefficient of determination (R²), calibration range, LLOQ and LLOD for superwarfarin stereoisomers in human plasma.

Superwarfarin	Coefficient of determination (R2)	Calibration range (ng/mL)	LLOQ (ng/mL)	LLOD (ng/mL)
cis-1 BDL	0.9993	1.43 – 91.54	1.43	0.72
cis-2 BDL	0.9994	1.57 – 100.40	1.57	0.78
trans-1 BDL	0.9993	1.03 – 33.03	1.03	0.52
trans-2 BDL	0.9992	0.95 – 30.40	0.95	0.48
cis-1 DFC	0.9997	0.87 – 221.52	0.87	0.43
cis-2 DFC	0.9997	0.92 – 236.20	0.92	0.46
trans-1 DFC	0.9998	1.18 – 303.15	1.18	0.59
trans-2 DFC	0.9998	1.03 – 263.13	1.03	0.51
cis-1 BDF	0.9996	1.39 – 355.60	1.39	0.69
cis-2 BDF	0.9997	1.66 – 424.22	1.66	0.83
trans-1 BDF	0.9996	2.55 – 651.98	2.55	1.27
trans-2 BDF	0.9996	2.41 – 616.16	2.41	1.20

3.3. Selectivity and matrix effect

Six lots of blank human plasma were analyzed, and no interfering signals were detected at the retention times of each superwarfarin stereoisomer (Figure 3). Plasma produced small but insignificant matrix effects. Specifically, SRM signals of BDF and BDL were diminished by 6 – 10% and 5 – 7%, respectively, while the signal for DFC was enhanced 3 – 10% (data not shown).

Figure 3.

Selective detection of superwarfarins BDL, DFC and BDF using positive ion electrospray HPLC-MS/MS. Blank human plasma (left column) was compared with blank plasma spiked with each superwarfarin stereoisomer at the LLOQ (right column).

3.4. Precision and accuracy

Intermediate precision, repeatability and accuracy were assessed at the LLOQ, low, medium and high concentrations for each superwarfarin stereoisomer. The intra-day repeatability (RSD) of 5 replicates at the LLOQ, low, medium and high concentrations for all stereoisomers were <8%, <10%, <8% and <6%, respectively, and the associated accuracies were between 98–113%, 97–101%, 93–106% and 98–105%, respectively. Inter-day precision (RSD) of 15 replicates at LLOQ, low, medium and high concentrations for all stereoisomers were <9%, <10%, <7% and <7%, respectively, with accuracies between 99–111%, 96–100%, 95–104% and 98–105%, respectively (Table 3).

Table 3.

Precision and accuracy of superwarfarin stereoisomers measured using HPLC-MS/MS.

	Intra-day (N = 5)				Inter-day (N = 15)
Superwarfarin	Spiked conc. (ng/mL)	Measured conc. (ng/mL)	Repeatability precision (%RSD)	Accuracy (%)	Measured conc. (ng/mL)	Repeatability precision (%RSD)	Accuracy (%)
cis-1 BDL	1.43	1.52	4.42	106.18	1.48	5.74	103.3
	4.29	4.33	3.81	100.81	4.29	3.67	100.1
	35.76	36.24	6.17	101.36	35.22	5.17	98.5
	71.51	71.74	3.47	100.32	72.66	4.72	101.6
cis-2 BDL	1.57	1.64	6.79	104.67	1.62	5.46	103.2
	4.71	4.70	2.57	99.79	4.64	2.73	98.6
	39.22	41.33	2.96	105.39	40.75	5.46	103.9
	78.44	78.17	3.19	99.66	79.22	5.89	101.0
trans-1 BDL	1.03	1.05	4.79	102.09	1.06	4.74	102.4
	3.10	3.06	3.65	98.95	3.06	4.06	98.8
	12.90	13.63	6.38	105.60	13.04	6.98	101.1
	25.81	27.13	5.54	105.11	26.80	5.18	103.8
trans-2 BDL	0.95	1.08	4.99	113.29	1.05	5.56	110.9
	2.85	2.86	3.64	100.25	2.83	4.28	99.2
	11.88	12.31	4.69	103.63	12.26	5.48	103.2
	23.75	23.51	2.07	98.98	23.59	4.18	99.3
cis-1 DFC	0.87	0.85	2.99	98.24	0.87	3.90	100.1
	2.60	2.57	2.42	99.17	2.56	2.05	98.5
	86.53	84.80	4.28	98.00	84.04	5.92	97.1
	173.06	180.89	4.27	104.52	181.37	5.00	104.8
cis-2 DFC	0.92	0.97	3.38	105.53	0.96	3.55	103.8
	2.77	2.76	2.93	99.86	2.75	2.28	99.3
	92.27	90.50	4.69	98.08	89.78	4.49	97.3
	184.53	186.58	3.53	101.11	188.01	4.09	101.9
trans-1 DFC	1.18	1.22	1.24	102.89	1.20	3.77	101.5
	3.55	3.55	9.57	99.86	3.48	9.83	97.9
	118.42	112.04	8.47	94.61	113.14	6.12	95.5
	236.84	232.95	5.82	98.36	233.46	5.44	98.6
trans-2 DFC	1.03	1.04	5.55	100.78	1.02	7.62	99.3
	3.08	2.98	2.45	96.71	2.98	2.85	96.5
	102.78	107.84	4.79	104.92	106.78	6.09	103.9
	205.57	202.69	3.88	98.60	203.33	4.03	98.9
cis-1 BDF	1.39	1.40	7.58	100.73	1.41	7.44	101.2
	4.17	4.13	6.50	99.10	4.11	5.16	98.7
	138.92	129.15	6.92	92.97	132.37	6.48	95.3
	277.84	279.52	2.55	100.61	282.32	3.88	101.6
cis-2 BDF	1.66	1.76	4.94	106.30	1.70	8.32	102.6
	4.97	4.98	6.39	100.19	4.79	6.92	96.4
	165.71	159.65	7.79	96.34	159.62	5.68	96.3
	331.43	328.77	2.46	99.20	339.45	6.96	102.4
trans-1 BDF	2.55	2.62	7.37	102.84	2.61	7.11	102.3
	7.64	7.68	4.60	100.46	7.45	5.66	97.5
	254.68	246.18	4.05	96.66	249.05	4.23	97.8
	509.36	498.97	2.87	97.96	497.56	4.96	97.7
trans-2 BDF	2.41	2.56	6.11	106.31	2.50	8.73	103.8
	7.22	7.20	3.42	99.74	7.20	2.45	99.7
	240.69	243.97	4.24	101.36	241.93	4.94	100.5
	481.38	494.64	3.28	102.76	485.07	4.33	100.8

3.5. Recovery.

A reported previously [39], the recoveries of BDL, DFC and BDF from human plasma at low, medium and high concentration were 86–96%, 87–95% and 81–98%, respectively.

3.6. Stability

Short-term (24 h at 25 °C and 4 °C) and long-term (60 days at −20 °C and −80°C) stabilities of BDL, DFC and BDF were reported previously [39]. Compared to freshly prepared samples, the observed loss in signal was <10% for each superwarfarin.

3.7. Application

The half-lives of each of the 4 enantiomers of BDL, DFC and BDF were evaluated in patient plasma using the new quantitative chiral HPLC-MS/MS method. Eight sets of patient plasma samples (3 females and 5 males with ages between 23 and 63), from individuals who inhaled synthetic cannabinoids tainted with BDL, DFC and BDL were analyzed [17,41,42] (Figure 4). The concentrations of BDL, DFC and BDF were measured in serial plasma specimens from each hospitalized patient (3 – 7 time points each), and the half-life of each superwarfarin stereoisomer was determined (Table S1). Due to the long half-lives of superwarfarins, the small number of samples, and relatively short duration of the serial blood draws, extrapolating half-lives contributed to inter-individual variation (Table S1). In a few cases, the half-lives could not be calculated due to the positive slope in concentration over time or had to be excluded as outliers (CI = 95%). Although cis-DFC diastereomers were detected, the absence of trans-DFC diastereomers from patient plasma indicated that these stereoisomers were eliminated much faster in humans (Table S2). No other significant differences between stereoisomers of each superwarfarin were observed.

Figure 4.

HPLC-MS/MS analysis of plasma samples show BDL, DFC and BDF in patients who had inhaled synthetic cannabinoid products contaminated with superwarfarin rodenticides. For case number 14, the concentration of each superwarfarin stereoisomer was as follows: (cis-1, cis-2, trans-1, trans-2): BDL (71, 14, 35, 17 ng/mL), DFC (18, 12, 0, 0 ng/mL) and BDF (563, 144, 311, 371 ng/mL). Note the absence of both trans-DFC enantiomers.

4. Discussion

The ability of a chiral column to resolve enantiomers is based on the differential intermolecular interactions between the chiral stationary phase and the stereoisomers. The greater the differences of interaction rates between the stereoisomers with the stationary phase, the better the selectivity of the chiral column. These interactions include hydrogen bonding, dipole-dipole, π-π stacking, and steric attraction and repulsion. The ridged structures of the four stereoisomers of BDF, DFC and BDL possess the functional groups capable of each of these types of intermolecular interactions.

With the exception of the Phenomenex Lux iCellulose-5 chiral column, the functionalized glucose-based and fructose-based chiral stationary phases from Phenomenex (Lux Cellulose-1, Lux Cellulose-2, Lux Cellulose-3, Lux Cellulose-4, Lux Amylose-1 and Lux iAmylose-1) and AZYP (CD Shell-RSP, Larihc Shell-P, Larihc CF6-RN and Larihc CF7-DMP) did not possess the selectivity required to separate the four stereoisomers of BDF, DFC and BDL. Therefore, each of the stereoisomers interacted with the chiral stationary phase at relatively the same rates resulting in the elution of the four superwarfarin stereoisomers as a single chromatographic peak. This was also the case for the AZYP glycopeptide-based chiral columns (Teico Shell, Vanco Shell, Nico Shell). The stationary phase polarity, steric accessibility to the asymmetric carbons of the stationary phase, and selective intermolecular interactions negatively affected their chromatographic separation.

The tris(3,5-dichlorophenylcarbamate) functionalized cellulose-based stationary phase of the Lux iCellulose-5 chiral column provided better selectivity to separate the four superwarfarin stereoisomers. This was attributed to the polarization of the aromatic ring by the chlorine substituents through electronegative induction. However, baseline resolution could not be achieved within a reasonable amount of time due to significant band broadening, which was due, at least in part, to the large particle size. Only the quinine-based AZYP Q-Shell chiral column had sufficient selectivity to achieve baseline separation of the four stereoisomers of BDF, DFC and BDL. Increased selectivity of the Q-Shell chiral column was due to the more accessible ion-exchange interactions between the amino groups of quinine and the alcohol groups of the coumarin moiety, as well as additional hydrogen bonding, and π-π stacking interactions. Optimization of the chromatographic conditions permitted rapid analysis and allowed for accurate quantitation of each stereoisomer of BDF, DFC and BDL.

Stereoselective pharmacokinetics are observed for many pharmaceuticals and pesticides such as (R)- and (S)-warfarin [49]. The clearance rate and half-life of (S)-warfarin are significantly lower than for (R)-warfarin [33,50,51]. This selectivity has also been observed for the diastereomeric pairs of DFC [27,29,36,39]. Here, we investigated the half-lives of BDL, DFC and BDF stereoisomers in humans using chiral chromatography (Figure 2 and Figure 4). The identification of one stereoisomer with rapid clearance and shorter half-life in humans, if still a potent inhibitor of VKORC1, might prove a safer alternative to the prolonged toxicity of currently commercially available superwarfarin racemates.

Plasma samples were obtained from hospitalized individuals who had smoked synthetic cannabinoid products contaminated with superwarfarins. Three superwarfarins, BDL, DFC and BDF, were detected in their plasma. This observation is noteworthy due to the fact that commercially available rodenticides are sold with only a single superwarfarin, at a concentration of 0.005% [52,53]. However, only recently have there been efforts to combine non-anticoagulant rodenticides with anticoagulant rodenticides, such as superwarfarins [25]. This suggests that these cannabinoid products were not accidentally contaminated by a single poison but were instead mixed with three separate superwarfarins.

Phoenix WinNonlin pharmacokinetic modeling software was used to determine the half-life of each superwarfarin stereoisomer based on their respective concentration versus time curves. To accurately determine the terminal half-life of superwarfarin stereoisomers, or for any xenobiotic, the concentration versus time slope requires samples to be acquired over three times the estimated half-life [54]. The half-life of xenobiotics can be estimated with fewer time points, one or two half-lives, but this can lead to an underestimation with a large amount of error.

In this half-life analysis, the number of blood draws over time was limited by the duration of patient hospitalization. Each case had between 3 – 7 blood samples drawn over a period of 2 – 7 days, which is less than required for proper calculation and therefore increased the error of the half-life estimation (Table S1). A few stereoisomer half-lives, for patient cases 14, 30 and 32, could not be approximated due to the positive slope fitted onto their respective concentration versus time curves. Additionally, some of the approximated half-lives, for cases 14 and 32, were determined to be outliers and were removed before statistical comparison.

The calculated half-lives of each stereoisomer (cis-1, cis-2, trans-1 and trans-2) of BDL, DFC and BDF were (37.3 ± 49.8%, 34.0 ± 65.5%, 56.0 ± 44.4%, 64.9 ± 56.3%), (278.0 ± 108.7%, 130 ± 83.7%, not detected, not detected) and (256.8 ± 83.1%, 186.1 ± 55.2%, 153.5 ± 64.9%, 162.2 ± 93.9%), respectively (Table S1). Statistical analysis showed that no significant differences were observed for the detected stereoisomers of each superwarfarin BDL, DFC and BDF (Table S2 and Figure S1). The absence of both trans-DFC diastereomers and the presence of cis-DFC diastereomers in patient plasma indicates that there is a significant difference between their respective elimination rates. Although not statistically significant, the stereoisomers of BDL and BDF with the shortest half-lives were cis-BDL (2) and trans-BDF (1) at 34.0 and 153.5 hours, respectively (Table S1). In agreement with the restrictions and regulations imposed by the US Environmental Protection Agency to mitigate rodenticide toxicity, our analysis suggests that cis-BDL (2), trans-DFC (1), trans-DFC (2) or trans-BDF (1) stereoisomers, the superwarfarin stereoisomers with the shortest half-lives, are suitable candidates for the next generation of anticoagulant rodenticides.

5. Conclusion

We have developed and validated a fast and stereoselective HPLC-MS/MS-based method to measure the stereoisomers of BDL, DFC and BDF in human plasma. Comparison of superwarfarin stereoisomer half-lives in human plasma from poisoning victims provided valuable insight into the persistence of each in humans but should be applied to a more complete sample set for greater statistical power. This method may be used to guide the development of the next generation of anticoagulant rodenticides that are effective against rodents but less persistent in humans.