TY - JOUR AU1 - Werneth, Madeleine AU2 - Pani, Jutta AU3 - Hofbauer, Ludwig AU4 - Pummer, Stefan AU5 - Weber, Maria-Theres AU6 - Pour, Georg AU7 - Kählig, Hanspeter AU8 - Mayer-Helm, Bernhard AU9 - Stepan, Herwig AB - Abstract The carcinogenic compound N-nitrososarcosine (NSAR) is found in foods and tobacco products, and its quantification is of great interest. Although the presence of two stereoisomers, E- and Z-NSAR, is well-known, individual investigation of the isomers has not been reported so far. The present study by liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) reveals that (i) the mass spectrometric responses of the isomers differ by a factor of approximately two and (ii) the isomer ratio is unstable in freshly prepared standard solutions. As a consequence, NSAR concentrations determined by LC–ESI–MS/MS are biased if those facts are not taken into account. The method described here overcomes the difficulty of stereospecific response by adjusting the isomer ratio and was applied to 100 tobacco products and fully validated for moist and dry snuff reference materials showing expanded measurement uncertainties of ~20% and limits of quantification of ~20 ng/g. Introduction N-nitrosamines, particularly tobacco-specific N-nitrosamines, are main contributors to the carcinogenicity of smokeless tobacco products, such as chewing tobacco (for example loose–leaf tobacco) or oral snuff (1). They evolve when amino compounds react with nitrite sources during aging and processing of the tobacco leaves (1). In general, three kinds of N-nitrosamines have been classified in smokeless tobacco with respect to their physical properties: tobacco-specific N-nitrosamines, volatile nitrosamines and (non-volatile) nitrosamino acids (1, 2). N-nitrososarcosine (NSAR) is of the latter class and is formed by nitrosation of the non-proteinogenic amino acid sarcosine. NSAR is among the least abundant N-nitrosamines in smokeless tobacco products with concentrations in the ng/g range (1, 2). Besides the low abundance of the target analyte in the highly complex tobacco matrix and besides its light-sensitivity and instability in aqueous solution, liquid chromatography–mass spectrometry (LC–MS) analysis of NSAR faces three major challenges: (i) The small molecule NSAR is highly polar because it bears an acid functionality. This property limits the liquid chromatographic retention on conventional reversed-phase (RP) stationary phases. The addition of ion-pairing reagents to the mobile phase facilitates the use of RP materials but interferes with electrospray ionization (ESI) and MS. (ii) Due to its acidic group, the yield of positive ions during electrospray ionization is low. Furthermore, because of the low molecular weight the ions appear in the low m/z range that is susceptible to high noise. (iii) Like other nitrosamines, NSAR bears partial double bonds that restrict rotation (3). Consequently, NSAR exists as two stereoisomers, termed E/Z-isomers that have different physical and chemical properties. Chow and Polo analyzed the conformational states of E- and Z-isomers of NSAR by NMR spectroscopy and observed that NSAR crystallizes with the Z-configuration and isomerizes to the E-configuration in solution with an isomer ratio of nearly 1:1 at equilibrium (4). The chemical structures of Z- and E-NSAR are shown in Figure 1. The International Agency for Research on Cancer (IARC) has classified NSAR as possibly carcinogenic to humans (group 2B), since a carcinogenic effect of NSAR was documented in mouse and rat (5, 6), but no information on the toxicology of the individual NSAR isomers is given (6). Figure 1 Open in new tabDownload slide Chemical structures of Z- and E-isomer of NSAR as proposed by (4). Figure 1 Open in new tabDownload slide Chemical structures of Z- and E-isomer of NSAR as proposed by (4). Several analytical methods for the determination of NSAR in different matrices have been described in the literature. For example, gas chromatography–thermal energy analyzer (GC–TEA) has been used for the determination of NSAR in malt and beer (7) or cigarette tobacco (8). Other gas chromatographic techniques, such as GC–flame ionization detector (9), GC–electron capture detector (10) and GC–MS (9, 10), have been proposed as well, although only for the analysis of standard solutions. All GC methods have the drawback that derivatization of the non-volatile NSAR is necessary. Moreover, the low selectivity of a TEA handicaps the distinction of co-eluting nitroso compounds. Although in GC only one peak is observed for NSAR due to the high interconversion rate at elevated temperatures (11), LC methods permit the separation of E- and Z-isomers. For example, LC coupled with an ultraviolet (UV) detector (11) and ion-pair RP LC coupled with a UV detector (12) or an ESI single quadrupole mass spectrometer in the negative ionization mode (13) were used to study the separation of N-nitrosamino acids and their E- and Z-isomers in standard solutions. Beyond doubt, when it comes to trace analysis of NSAR in real samples, LC–tandem mass spectrometry (MS/MS) is state-of-the-art technology and the method of choice. Few LC–MS/MS methods based on RP chromatography and different ionization techniques have been developed to determine NSAR, e.g., in tobacco and smokeless tobacco products (14, 15) and in meat (16, 17). Comprehensive multi-analyte methods are impressive and offer the advantage to analyze multiple targets within a single run (15, 16). However, they cannot always meet the requirements to quantify trace amounts of an individual analyte in a certain, complex matrix. The method presented here is customized to the needs of the determination of NSAR in tobacco-related matrices, which is achieved by the combination of hydrophilic interaction liquid chromatography (HILIC) with negative mode electrospray ionization. The presented method was validated and successfully applied to 100 tobacco products. It revealed different ESI–MS responses for the individual NSAR isomers and was used to investigate their mass spectrometric behavior as well as their isomerization characteristics. Experimental Chemicals and reagents Acetonitrile (gradient grade for LC, ≥99.9%) and formic acid (p.A., 98–100%,) were purchased from Merck (Darmstadt, Germany) and ammonium formate (for MS, ≥99.9%) from Sigma Aldrich Chemie GmbH (Steinheim, Germany). Water was treated using a purification system (Millipore Advantage A10, Merck Millipore, Darmstadt, Germany). Ethyl formate (≥98%) was obtained from Karl Roth GmbH (Karlsruhe, Germany). Standards of NSAR (98%) and NSAR-d3 (chemical purity 97%, isotopic purity 99.8%) were purchased from Toronto Research Chemicals Inc. (Toronto, Canada). Stock solutions (0.5 mg/mL) of NSAR and NSAR-d3 in acetone (p.A. from Merck) were diluted with mobile phase B (5 mM ammonium formate and 0.1% (v/v) formic acid in 95/5% (v/v) acetonitrile/water) to give spiking solutions with concentrations of 5 μg/mL. Kentucky Reference Cigarettes 3R4F and 1R5F were obtained from the Tobacco and Health Research Institute (University of Kentucky, Lexington, KY, USA), CORESTA monitor cigarettes CM6 and CM8 were obtained from the Cooperation Centre for Scientific Research Relative to Tobacco (Paris, France) and smokeless tobacco reference products CRP1 (snus smokeless tobacco), CRP2 (moist snuff), CRP3 (dry snuff) and CRP4 (loose leaf chewing tobacco) were obtained from the Tobacco Analytical Services Laboratory (North Carolina State University, Raleigh, NC, USA). A total of 92 real-life samples were snus, tobacco, cigarette tobacco, waterpipe tobacco and stems samples from different suppliers. Sample preparation Moist and dry snuff did not require any special pre-treatment. Snus pouches (e.g., CRP1), leaves of chewing tobacco (e.g., CRP4) and waterpipe tobacco were cut into pieces to increase sample homogeneity. Cigarette tobacco, tobacco and stems were ground. Sample preparation was performed following Wu et al. (14). Briefly, 2.0 g of sample were spiked with 75 μL of NSAR-d3 spiking solution and extracted with 25 mL of 2% formic acid in water (v/v) in a 50 mL polypropylene tube and agitated for 45 min using a test tube shaker at 2,000 rpm. Afterward, the sample was centrifuged (2,370 × g, 5 min, room temperature) and 10 mL of the supernatant were loaded onto a 10 mL solid-supported liquid extraction cartridge (Chem Elut, Agilent Technologies, Waldbronn, Germany). After 15 min of equilibration, the analyte was eluted from the cartridge with 2 × 20 mL of ethyl formate. The eluate (~30 mL) was then evaporated to dryness under a nitrogen stream in a waterbath at 50°C and subsequently reconstituted with 1 mL of mobile phase B (5 mM ammonium formate and 0.1% (v/v) formic acid in 95/5% (v/v) acetonitrile/water). Finally, the solution was filtered (syringe-driven filter units, 0.45 μm, PTFE, Millex-LCR from Merck Millipore, Darmstadt, Germany) and transferred to an autosampler vial. For preparation of the external calibration standard as well as the standard solutions used for the isomer ratio investigations, 80 μL of the NSAR spiking solution and 12 μL of the NSAR-d3 spiking solution were added to 3.8 mL of mobile phase B resulting in final concentrations of 100 ng/mL NSAR and 150 ng/mL NSAR-d3. A cabinet dryer (UF55 from Memmert, Schwabach, Germany) was used for heat treatments. The external calibration standard was incubated at 60°C for 90 min prior to injection. To investigate the isomer ratio of real-life samples, CRP3 and a stems sample were analyzed with the LC separation method after modified sample preparation: To diminish isomer conversion due to elevated temperatures during sample preparation, solvent evaporation was performed at room temperature under a nitrogen stream using a TurboVap® II from Biotage, Uppsala, Sweden. LC–MS/MS analysis The chromatographic system consisted of 1,290 Infinity components (Agilent Technologies): a binary pump (G4220A) with an integrated degasser, an autosampler (G4226A) equipped with a cooling device (G1330B) and a column compartment (G1316C). The separation was carried out on an Obelisc N column from SIELC Technologies, Wheeling, IL, USA (2.1 × 150 mm, 5 μm particles, 10 Å pores) protected by an Obelisc N 2.1 × G mm guard column, both temperature-controlled at 35°C. Mobile phase A (5 mM ammonium formate and 0.1% (v/v) formic acid in 95/5% (v/v) water/acetonitrile) and mobile phase B (5 mM ammonium formate and 0.1% (v/v) formic acid in 95/5% (v/v) acetonitrile/water) were filtered through 0.45 μm polyamide filters (Sartorius Stedim Biotech GmbH, Göttingen, Germany) prior to use. Two different gradients at a flow rate of 0.4 mL/min were applied depending on whether a separation of the isomers was intended or not. For co-elution of the isomers the following gradient was used: 0 min, 100% B; 2.2 min, 100% B; 6 min, 10% B; 7 min, 10% B; 7.1 min, 100% B and 37 min, 100% B. For separation of the isomers a different gradient was used: 0 min, 100% B; 3 min, 100% B; 8 min, 10% B; 9 min, 10% B; 9.1 min, 100% B and 14 min, 100% B. Injection volume was set to 10 μL, autosampler tray temperature to 8°C. The LC system was coupled to a SCIEX Triple Quad 5,500 mass spectrometer (SCIEX, Framingham, MA, USA) operating in the multiple reaction monitoring (MRM) mode. The Turbo V™ ion source was used for negative ESI with an ion spray voltage of −4,500 V and a source temperature of 450°C. Nitrogen gas flows were established using flow injection analysis and set as follows: nebulizer gas (gas 1): 50 psi, heating gas (gas 2): 60 psi, curtain gas: 25 psi, collision gas: 6 psi. The compound-dependent MRM parameters for the three monitored transitions (optimized by syringe-driven direct infusion) are listed in Table I. The investigation of the ESI–MS/MS behavior of the individual isomers was performed following the protocol of Mayer-Helm et al. (18). Briefly, the isomers of NSAR were separated via LC and full scan spectra were recorded for both isomers. Then the declustering potential was ramped in 10 V steps (from −10 to −100 V) from LC run to run in the selected ion monitoring (SIM) mode. Then product ion spectra were recorded for both isomers and product ions were selected to build MRM transitions. Finally, the collision energy was ramped in 2.5 eV steps (from 5 to 25 eV) during consecutive LC runs in the MRM mode. Validation For method validation, the following parameters were assessed for the smokeless tobacco reference products CRP2 and/or CRP3: interday repeatability, recovery, limit of detection (LOD), limit of quantification (LOQ), linearity, selectivity and measurement uncertainty. The interday repeatability was tested by analyzing CRP2 and CRP3 on seven different days. The recovery of NSAR was determined by standard addition on six different days and was calculated as follows: peak area of spiked matrix minus peak area of unspiked matrix divided by peak area of standard solution multiplied by 100 (in %). The recovery of NSAR-d3 was calculated as peak area of NSAR-d3 spiked matrix divided by peak area of standard solution multiplied by 100 (in %). In order to determine whether the analyte was lost during sample preparation or during measurement due to signal suppression, CRP3 extracts were spiked with NSAR standards before and after sample preparation. LOD and LOQ corresponded to the concentrations with signal to noise ratios (S/N) of 3:1 and 10:1, respectively, based on the signal of the qualifier transition in order to ensure identity confirmation (19, 20). For linearity assessment, CRP2 extracts were UV-treated at a wavelength of 312 nm for 18 h to generate a matrix without detectable residual analyte and were subsequently spiked with NSAR concentrations of 12.5, 18.8, 25.0, 62.5 and 125.0 ng/g. Selectivity was assessed based on relative intensity (area of qualifier transition divided by area of quantifier transition) and retention time (20). Measurement uncertainty was calculated as follows: random error (precision as relative standard deviation (RSD), n = 4) multiplied by an expanded coverage factor k = 3. Because of the lack of reference values, the systematic error has not been taken into account for measurement uncertainty calculation. Further validation parameters were assessed in standard solution and/or in different real-life samples: stability of stock solutions and reproducibility of the two LC methods (5 vs. 30 min equilibration time). The latter parameter was tested by determining the interday reproducibility of retention times in quintuplicate on four different days. 1H NMR spectroscopy For the quantification of the isomer ratio at equilibrium, NSAR and NSAR-d3 were dissolved in acetone, incubated at 60°C for 90 min, evaporated to dryness applying a nitrogen stream and redissolved in acetone-d6. The samples were measured in 5 mm outer diameter tubes (Wilmad, Buena, NJ, USA) at a temperature of 25°C. 1H NMR spectra were recorded at a frequency of 600.13 MHz on a Bruker Advance III 600 NMR spectrometer (Bruker BioSpin, Ettlingen, Germany) equipped with a 5 mm broadband observe probe. All chemical shifts were referenced to the residual solvent signal of acetone (δ = 2.05 ppm). Signals were assigned according to Chow et al. (4). The singlets at 3.08 and 3.88 ppm corresponded to the CH3 groups of E- and Z-NSAR, respectively. The singlets at 5.06 and 4.35 ppm corresponded to the CH2 groups of E- and Z-NSAR, respectively. Quantification of the isomer ratio was based on the signal areas after deconvolution. Table I Optimized MRM Parameters for the Detection of NSAR and NSAR-d3 . Precursor ion, m/z . Product ion, m/z . Declustering potential (V) . Entrance potential (V) . Collision energy (eV) . Collision cell exit potential (V) . Quantifier 117.0 32.1 −35 −10 10 −5 Qualifier 117.0 30.0 −35 −10 16 −5 NSAR-d3 120.0 32.1 −35 −10 10 −5 . Precursor ion, m/z . Product ion, m/z . Declustering potential (V) . Entrance potential (V) . Collision energy (eV) . Collision cell exit potential (V) . Quantifier 117.0 32.1 −35 −10 10 −5 Qualifier 117.0 30.0 −35 −10 16 −5 NSAR-d3 120.0 32.1 −35 −10 10 −5 Open in new tab Table I Optimized MRM Parameters for the Detection of NSAR and NSAR-d3 . Precursor ion, m/z . Product ion, m/z . Declustering potential (V) . Entrance potential (V) . Collision energy (eV) . Collision cell exit potential (V) . Quantifier 117.0 32.1 −35 −10 10 −5 Qualifier 117.0 30.0 −35 −10 16 −5 NSAR-d3 120.0 32.1 −35 −10 10 −5 . Precursor ion, m/z . Product ion, m/z . Declustering potential (V) . Entrance potential (V) . Collision energy (eV) . Collision cell exit potential (V) . Quantifier 117.0 32.1 −35 −10 10 −5 Qualifier 117.0 30.0 −35 −10 16 −5 NSAR-d3 120.0 32.1 −35 −10 10 −5 Open in new tab Results Chromatography Two RP columns were tested for their ability to retain NSAR: a ZORBAX Eclipse XDB-C18 (Agilent) and a Synergi Hydro-RP (Phenomenex). It turned out that NSAR eluted close to the dead time on both columns and that retention was only slightly improved by reducing column temperature or optimizing the eluent starting conditions. Since the use of ion-pairing reagents is unfavorable for ESI–MS/MS, the method was based on a HILIC application note using an Obelisc N column as stationary phase and ammonium formate as mobile phase modifier (21). The finally selected eluent composition provided reasonable retention of NSAR (Figure 2). An equilibration time of 30 min was required to ensure stable conditions for acceptable retention of NSAR isomers (co-eluting at 4.4 min, see Figure 2a). After an equilibration time of 5 min, the interactions with the NSAR isomers were generally weaker (average retention time of 2.4 min) and Z-NSAR eluted prior to E-NSAR (see Figure 2b). So, both co-elution and separation of NSAR isomers solely depended on the duration of the equilibration time of the column prior to injection without altering the mobile phase or gradient itself. Compared to RP chromatography, HILIC is generally less robust against changes of eluent composition and requires longer equilibration times (typically 15–20 column volumes) (22). The importance to specify the re-equilibration times for gradient HILIC methods has been recently outlined (23, 24). Figure 2 Open in new tabDownload slide MRM chromatograms (m/z 117.0 → 32.1) of a 100 ng/mL NSAR standard solution: (a) co-elution at 4.4 min after 30 min equilibration time, (b) separation of the two isomers after 5 min equilibration time. Figure 2 Open in new tabDownload slide MRM chromatograms (m/z 117.0 → 32.1) of a 100 ng/mL NSAR standard solution: (a) co-elution at 4.4 min after 30 min equilibration time, (b) separation of the two isomers after 5 min equilibration time. Both LC methods provided consistent retention times with high intraday and interday precision, determined in quintuplicate on four different days (RSD < 1% for both methods), see Table II. Table II Reproducibility of Retention Times of Both Methods . Intraday precision . Interday precision (k = 4) . Method . Day 1 . Day 2 . Day 3 . Day 4 . . Co-elution  NSAR   Retention times average (n = 5) 4.366 4.446 4.446 4.454 4.428   RSD (%) 0.18 0.14 0.12 0.10 0.94 Separation  Z-NSAR   Retention times average (n = 5) 2.315 2.314 2.314 2.350 2.323   RSD (%) 0.13 0.08 0.24 0.14 0.77  E-NSAR   Retention times average (n = 5) 2.507 2.510 2.516 2.539 2.518   RSD (%) 0.12 0.09 0.22 0.24 0.58 . Intraday precision . Interday precision (k = 4) . Method . Day 1 . Day 2 . Day 3 . Day 4 . . Co-elution  NSAR   Retention times average (n = 5) 4.366 4.446 4.446 4.454 4.428   RSD (%) 0.18 0.14 0.12 0.10 0.94 Separation  Z-NSAR   Retention times average (n = 5) 2.315 2.314 2.314 2.350 2.323   RSD (%) 0.13 0.08 0.24 0.14 0.77  E-NSAR   Retention times average (n = 5) 2.507 2.510 2.516 2.539 2.518   RSD (%) 0.12 0.09 0.22 0.24 0.58 Open in new tab Table II Reproducibility of Retention Times of Both Methods . Intraday precision . Interday precision (k = 4) . Method . Day 1 . Day 2 . Day 3 . Day 4 . . Co-elution  NSAR   Retention times average (n = 5) 4.366 4.446 4.446 4.454 4.428   RSD (%) 0.18 0.14 0.12 0.10 0.94 Separation  Z-NSAR   Retention times average (n = 5) 2.315 2.314 2.314 2.350 2.323   RSD (%) 0.13 0.08 0.24 0.14 0.77  E-NSAR   Retention times average (n = 5) 2.507 2.510 2.516 2.539 2.518   RSD (%) 0.12 0.09 0.22 0.24 0.58 . Intraday precision . Interday precision (k = 4) . Method . Day 1 . Day 2 . Day 3 . Day 4 . . Co-elution  NSAR   Retention times average (n = 5) 4.366 4.446 4.446 4.454 4.428   RSD (%) 0.18 0.14 0.12 0.10 0.94 Separation  Z-NSAR   Retention times average (n = 5) 2.315 2.314 2.314 2.350 2.323   RSD (%) 0.13 0.08 0.24 0.14 0.77  E-NSAR   Retention times average (n = 5) 2.507 2.510 2.516 2.539 2.518   RSD (%) 0.12 0.09 0.22 0.24 0.58 Open in new tab The co-elution method (30 min equilibration) provided adequate retention, which is a prerequisite for the separation from matrix peaks and is essential for trace analysis in the highly complex tobacco matrix. Therefore, the co-elution method was used for NSAR quantification in real-life samples, as demonstrated in Figure 3a, continuous line. To further illustrate the applicability, a chromatogram of a blank matrix (tobacco sample without detectable analyte) is presented for comparison in Figure 3a, dotted line. The isomer separation method (5 min equilibration) provided reasonable chromatographic separation of the isomers and was used for the investigation of the isomer ratio and the ESI–MS/MS behavior of NSAR isomers in standard solution. Figure 3 Open in new tabDownload slide MRM chromatograms of the (a) quantifier and (b) qualifier transition of a CRP2 extract using the LC method for co-elution (continuous lines). The dotted line in (a) shows the quantifier chromatogram of a tobacco sample without detectable analyte (an intensity offset of −250 cps was chosen for optimized illustration). Figure 3 Open in new tabDownload slide MRM chromatograms of the (a) quantifier and (b) qualifier transition of a CRP2 extract using the LC method for co-elution (continuous lines). The dotted line in (a) shows the quantifier chromatogram of a tobacco sample without detectable analyte (an intensity offset of −250 cps was chosen for optimized illustration). Mass spectrometry For development of the mass spectrometric detection, four MRM transitions of the precursor ion m/z 117.0 ([M-H]−) were tested for their applicability to tobacco samples (in brackets: abundance relative to the quantifier transition (m/z 117.0 → 32.0) in standard solution): m/z 117.0 → 73.1 (235%), 117.0 → 32.1 (100%), 117.0 → 75.0 (12%) and 117.0 → 30.0 (2.7%). The transition to the most abundant product ion (m/z 73.1, used by Wu et al. (14)), formed by the unspecific loss of CO2, turned out to be unfavorable in matrix due to a high baseline and interfering matrix background. The transition to the second most abundant product ion (m/z 32.1, H2NO−) proved to be useful for the determination of NSAR in tobacco samples since the target peak was well separated from matrix peaks (see Figure 3a, continuous line) and was therefore selected as quantifier. The transition m/z 117.0 → 75.1 was substantially interfered, whereas the transition m/z 117.0 → 30.0 (NO−) offered a low and undisturbed baseline (see Figure 3b). Hence, the latter transition was selected as qualifier. Isomer ratio investigations of NSAR by LC–MS/MS and NMR spectroscopy First, standard solutions of different ages were analyzed using the isomer separation method and the peak area percentages of the isomers were assessed revealing the following two observations: (i) The isomer ratio was not stable: the first eluting isomer decreased with the age of the standard, whereas the second eluting isomer increased. Based on the observation of Chow et al. (4) (NSAR crystallizes with the Z-configuration and isomerizes to the E-configuration in solution), the first peak was inferred to be Z-NSAR and the second peak to be E-NSAR (Figure 2b). (ii) Not only the peak area of the E-isomer but also the sum of both isomer peak areas increased with the age of the solutions implicating that the ESI–MS/MS responses of the isomers are different. The peak area of E-NSAR increased approximately twice as much as the peak area of Z-NSAR decreased. That means that the ESI–MS/MS response of E-NSAR was enhanced by a factor of about two compared to Z-NSAR. As both isomers eluted at almost the same mobile phase composition, a response difference evolving from different electrospray conditions could be excluded (25). In a next step, standard solutions after different heat treatments as well as CRP3 extracts were analyzed (Figure 4) showing the following: (i) Isomeric equilibration in solution was reached with a peak area percentage (of the peak area sum of both isomers) of E-NSAR of 72%. (ii) Isomeric equilibration was completed after 3.8 d at 25°C, after 5 h at 50°C and after 90 min at 60°C. (iii) The peak area percentage of E-NSAR in CRP3 extracts (spiked and non-spiked) was 75%. Figure 4 Open in new tabDownload slide Isomeric equilibration of NSAR in standard solution at different temperatures and comparison with the isomer ratio in a CRP3 extract. Isomer ratio is indicated as peak area percentage of E-NSAR (%). Figure 4 Open in new tabDownload slide Isomeric equilibration of NSAR in standard solution at different temperatures and comparison with the isomer ratio in a CRP3 extract. Isomer ratio is indicated as peak area percentage of E-NSAR (%). In order to quantify the MS response difference of the isomers, the isomer ratio at equilibrium was determined by NMR spectroscopy. The isomer ratio was calculated from the proton spectrum of NSAR as 1:1.085 (average, n = 2) and was compared with the peak area ratio of the LC–MS/MS measurements (1:2.36, average, n = 11). Thus, the intensity difference of E-NSAR compared to Z-NSAR can be described by a factor of 2.2 with an overall RSD of 5%. Also, the isomer ratio of the internal standard NSAR-d3 was investigated by LC–ESI–MS/MS and 1H NMR spectroscopy. NSAR-d3 showed the same behavior as NSAR regarding isomeric equilibration, isomer ratio at equilibrium and ESI–MS/MS response differences (data not shown). ESI–MS/MS behavior of NSAR isomers In order to find the source of the different ESI–MS/MS response of the isomers, investigation of the individual isomers was performed using the LC separation method. First, full scan spectra of the isomers were compared showing no qualitative difference: none of the isomers produced a certain adduct instead of the deprotonated molecule. The two most abundant ions for both isomers were the deprotonated molecule [M-H]− at m/z 117.0 and the deprotonated dimer [2 M-H]− at m/z 235.0. In order to quantify the intensity difference of the precursor ion m/z 117.0 and to assess possibly different declustering potential optima, the declustering potential was ramped from −10 to −100 V for both isomers at equilibrium (isomer ratio ~1:1) showing that both isomers had the same optimum declustering potential at approximately −35 V. However, the absolute intensity differed significantly by a factor of about two (in favor of the E-isomer, see Figure 5). Figure 5 Open in new tabDownload slide Dependence of the MS response of NSAR isomers at m/z 117.0 ([M-H]−) on the declustering potential in the SIM mode. Analyte concentration 300 ng/mL. Isomer ratio at equilibrium (~1:1). Figure 5 Open in new tabDownload slide Dependence of the MS response of NSAR isomers at m/z 117.0 ([M-H]−) on the declustering potential in the SIM mode. Analyte concentration 300 ng/mL. Isomer ratio at equilibrium (~1:1). In a next step, the collision energies of the four MRM transitions (m/z 117.0 → 73.1, 117.0 → 32.1, 117.0 → 75.1 and 117.0 → 30.0) were ramped from 5 to 25 eV demonstrating that the intensities differed constantly by a factor of about two (Figure 6a–d). This means that collision energy optima and relative intensities were very similar for both isomers; they differed mainly in their absolute peak intensities, caused by the differently abundant precursor ion. Figure 6 Open in new tabDownload slide MS response of NSAR isomers versus collision energy in the MRM mode for four different transitions. Declustering potential −35 V. Analyte concentration 100 ng/mL. Isomer ratio at equilibrium (~1:1). Figure 6 Open in new tabDownload slide MS response of NSAR isomers versus collision energy in the MRM mode for four different transitions. Declustering potential −35 V. Analyte concentration 100 ng/mL. Isomer ratio at equilibrium (~1:1). Since the response difference was already observed for the precursor ion, it can be concluded that the absolute intensity difference was generated at the beginning of the ESI–MS/MS process, as the E-isomer was more efficiently ionized than the Z-isomer. Quantification of NSAR in tobacco and smokeless tobacco products Regardless of the LC method (co-elution or separation) a calibration standard with known and stable isomer ratio is essential for unbiased quantification. Naturally, a calibration standard with isomers in equilibrium meets these demands best. As shown in Figure 4 and as mentioned above, isomeric equilibration in standard solution was completed within 90 min when the standard solution was incubated at 60°C. Since NSAR did not thermally degrade at this temperature, the incubated standard solution was chosen as external calibration standard as described in the experimental section. As discovered, the 1:1 isomer ratio of NSAR in the incubated standard was consistent with the isomer ratio in real-life samples. If the calibration standard and sample have the same isomer ratio, it is not necessary to correct for an isomer ratio difference. Hence, both LC methods (co-elution and separation) can be applied for quantification. In order to prove conformity, NSAR was determined in three smokeless tobacco products with the co-elution as well as the separation method with agreements of 87–98% (data not shown). Since the co-elution method offered better matrix separation and higher sensitivity, it was chosen for sample screening and method validation. NSAR concentrations were determined in eight tobacco and smokeless tobacco reference products (CRP1,