Development and validation of a method for quantification of two tobacco-specific nitrosamines in indoor air

A sensitive and accurate method for the quantification of 1 -Demethyl-1 -nitrosonicotine (NNN) and 4-(methylnitrosamino)-1-(3-Pyridyl)-1-butanone (NNK) in indoor air was developed and validated. To this aim, a novel approach for the collection of two tobacco-specific nitrosamines, using silica sorbent cartridges followed by simplified sample preparation and isotope dilution liquid chromatography/electrospray ionization tandem mass spectrometry, was applied.

Journal of Chromatography A, 1580 (2018) 90–99 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Development and validation of a method for quantification of two tobacco-specific nitrosamines in indoor air María Gómez Lueso a , Maya I Mitova a,∗ , Nicolas Mottier a,b , Mathieu Schaller a , Michel Rotach a , Catherine G Goujon-Ginglinger a a b PMI R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, CH-2000 Neuchâtel, Switzerland Service de la Consommation et des Affaires Vétérinaires, Chemin des Boveresses 155, 1066 Epalinges, Switzerland a r t i c l e i n f o Article history: Received 13 July 2018 Received in revised form 28 September 2018 Accepted 17 October 2018 Available online 23 October 2018 Keywords: Tobacco-specific nitrosamines Validation Accuracy profile Environmental aerosol Tobacco heating system e-Cigarette a b s t r a c t A sensitive and accurate method for the quantification of -Demethyl-1 -nitrosonicotine (NNN) and 4-(methylnitrosamino)-1-(3-Pyridyl)-1-butanone (NNK) in indoor air was developed and validated To this aim, a novel approach for the collection of two tobacco-specific nitrosamines, using silica sorbent cartridges followed by simplified sample preparation and isotope dilution liquid chromatography/electrospray ionization tandem mass spectrometry, was applied This procedure led to a substantial improvement in terms of sensitivity and sample throughput as compared with methods using conventional trapping For the validation, a matrix-based approach using an accuracy profile procedure was selected The evaluated matrices were background air samples, environmental aerosols of a heat-not® burn tobacco product (Tobacco Heating System [THS] 2.2, commercialized under the brand IQOS ), a ® rechargeable electronic cigarette (Solaris ), and the environmental tobacco smoke (ETS) of a conven® tional cigarette (Marlboro Gold ) The method showed excellent recoveries, sensitivity, and precision The limits of detection of the method for NNN and NNK were 0.0108 ng/m3 and 0.0136 ng/m3 , respectively The calibration range of the instrument spanned 0.2–60 ng/mL The calculated lower working range limit (LWRL) of the method for NNN was 0.126 ng/m3 , and the LWRL for NNK was 0.195 ng/m3 The method was applied to evaluate surrogate environmental aerosols generated using smoking machines This model is reliable but gives a large overestimation of the possible impact of THS 2.2 and e-cigarettes on indoor air, because the retention of NNN and NNK in the body of the consumers is not taken into account As a consequence, the values reported not reflect a real-life setting The contents of the two target compounds in the surrogate environmental aerosols were 0.0830 ± 0.0153 ng/m3 of NNN and 0.0653 ± 0.0138 ng/m3 of NNK for THS 2.2, 0.0561 ± 0.0296 ng/m3 of NNN for e-cigarettes, and 0.816 ± 0.109 ng/m3 of NNN and 4.13 ± 1.04 ng/m3 NNK for cigarettes These values correspond to 10% of the measured ETS concentration for NNN in environmental aerosols of THS 2.2 and 7% for those of e-cigarettes For NNK, the value for the environmental aerosol of THS 2.2 was 2% of the ETS value © 2018 PMI R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, 2000 Neuchâtel, Switzerland Published by Elsevier B.V This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/) Introduction Tobacco-specific nitrosamines (TSNA) are carcinogens known to be specifically associated with tobacco, tobacco smoke, and related nicotine-containing products [1] In 1964, demethyl-1 -nitrosonicotine; 1-nitroso-2-(3-pyridyl)pyrrolidine; N-nitrosonornicotine (NNN) was proven to cause pulmonary cancer in mice, as was 4-(methylnitrosamino)-1-(3-Pyridyl)- ∗ Corresponding author E-mail address: Maya.Mitova@pmi.com (M.I Mitova) 1-butanone; 4-(N-methyl-N-nitrosamino)-1-(3-pyridyl)-1butanone; nicotine-derived nitrosamine ketone (NNK) in 1980 [2] Further investigations demonstrated that both NNN and NNK are carcinogens inducing several types of cancer in laboratory animals, with NNK being more active than NNN [3,4] Both compounds are included on the U.S Food and Drug Administration (FDA) list of harmful and potentially harmful constituents in tobacco products and tobacco smoke [5] and are classified as carcinogens of Group by the International Agency for Research on Cancer [6] TSNAs are present at trace levels in freshly harvested tobacco; however, their concentration might vary depending on the type of tobacco and the fertilizers used during the tobacco plant grow- https://doi.org/10.1016/j.chroma.2018.10.037 0021-9673/© 2018 PMI R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, 2000 Neuchâtel, Switzerland Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 ing [7] NNN and NNK are mainly formed by nicotine nitrosation, although NNN can also be generated by nornicotine nitrosation [3,4] This process occurs mostly during the processing, curing, and storage of tobacco and tobacco products [3,7] In mainstream cigarette smoke, NNK and NNN derive partially from the distillation of these nitrosamines, which are pre-formed in the tobacco, and NNK is also the result of thermal release of the matrix-bound form, while another fraction is pyrosynthesized by nitrosation of the respective alkaloid precursors, possibly with nitrogen oxides originating from the nitrate, present in high concentrations in some tobacco types [3,8,9] NNK and NNN are also present in sidestream smoke, and their yields are at the same level or two to five times higher than those found in mainstream smoke [10,11] The formation of NNK is favored during the smoldering of cigarettes when sidestream smoke is generated [10] As a consequence, indoor air enriched with environmental tobacco smoke (ETS), defined as an aged and diluted mixture of exhaled smoke and sidestream smoke, contains NNN and NNK Published data on ETS in real-life and simulated environments indicate concentrations of both NNN and NNK in the low ng/m3 range (NNN: not detected − 23 ng/m3 ; NNK 1–29 ng/m3 ) [10–13] Studies of TSNA content in tobacco leaf [14,15] and in mainstream [14,16–19] and sidestream [20] cigarette smoke have been conducted over the years using different methodologies However, more recently, the liquid chromatography coupled to tandem mass spectrometry (LC–MS/MS) technique has become the reference methodology for TSNA analysis [21] An interlaboratory comparison of the performance of several methods for the quantification of TSNAs in mainstream smoke was published [19], but overall, few publications about TSNA content in ETS exist [11–13,22,23] In recent years, the impact of new products, such as electronic cigarettes [24–30] or heated tobacco systems [13,31–34], on indoor air quality has been evaluated Tricker et al [13] reported NNN concentrations at background levels (0.250 ng/m3 ), while NNK concentrations were in the range of 0.300–0.691 ng/m3 (background levels 0.300–0.602 ng/m3 ) during indoor use in simulated “Office” and “Hospitality” environments of a product developed in the 1980s The detected presence of TSNAs in background air was not explainable and might have been due to cross-contamination NNN and NNK were reported in the exhaled breath of e-cigarette users who vaped e-liquids that had been contaminated with TSNAs [35] However, to the best of our knowledge, NNN and NNK have not been investigated in the environmental aerosols of e-cigarettes Furthermore, and again to the best of our knowledge, since 1992 [12] there have been no publications reporting on improvements in TSNA trapping and analyses in air samples Recent publications on the quantification of airborne TSNAs describe approaches which applied procedures developed for mainstream smoke analyses [13,24,34,36] As a consequence, methods with LLOQ above 0.3 ␮g/m3 have been reported [24,34] Other publications describe approaches where the lowering of the reporting limits of the methods was achieved by laborious sample preparation procedures [13,36] As the concentrations of NNN and NNK in the environmental aerosols of heat-not-burn products and e-cigarettes are foreseen to be much lower than those in ETS [13,31–34,37,38], a new method aiming at improving both the sensitivity and sample throughput was developed in order to ensure quantification of NNN and NNK in air in a reliable manner A validation applying accuracy profiles was undertaken to allow rigorous evaluation of the method performance and any possible matrix effect on the quantification of the target compounds For the development and the validation of the method, the samples were collected in an environmentally controlled room Surrogate environmental aerosols and ETS were generated with smoking machines to improve reproducibility between experiments 91 Material and methods 2.1 Chemicals The following compounds were purchased from Sigma-Aldrich: NNN certified solution (1 mg/mL in methanol), NNK certified solution (1 mg/mL in methanol), tetrahydrofuran (HPLC grade), water ® with 0.1% formic acid CHROMASOLV (LC–MS grade), methanol CHROMASOLV (LC–MS grade), formic acid (eluent additive for LC–MS) and ethyl acetate The following compounds were purchased from Chemie Brunschwig AG: 4-(methylnitrosamino)-1-(3-pyridyl-D4 )-1-butanone (NNK-D4 , 0.1 mg/mL in methanol) and rac N’-NitrosonornicotineD4 (NNN-D4 , 0.1 mg/mL in methanol) The Sep-Pak Silica 690 mg sorbent cartridges were purchased from Waters 2.2 Test items For the validation of the TSNA method, four matrices were generated Ambient air of an empty, environmentally controlled room without consumption of any product was used as the background matrix Ambient air enriched with the mainstream aerosol of Tobacco Heating System (THS) 2.2 (marketed under the IQOS brand ® name) or a cig-a-like e-cigarette (marketed under the Solaris ® brand name in Spain and under the MarkTen brand name in the U.S.) were used as surrogate environmental aerosols of a heat-notburn product and e-cigarette, respectively Regular THS 2.2 was used for the experiments A detailed description of the THS 2.2 (Fig S1) has been presented by Smith et al [39] The Solaris KS type is an e-cigarette with a cartomizer and a rechargeable battery of 90 mA The cartomizer contains 0.4 mL of a tobacco-flavored liquid consisting of 20.3 mg/mL nicotine (Fig S1) The cigarettes used for generation of the surrogate ETS (aged and diluted sidestream smoke) were Marlboro Gold retailed on the Swiss market (characterized by mg tar, 0.5 mg nicotine, and mg carbon monoxide (CO) under International Organization for Standardization testing conditions) The Marlboro Gold cigarettes and THS 2.2 were manufactured by Philip Morris Products S.A, Neuchâtel, Switzerland The ® Solaris items were manufactured by Numark LLC, Richmond, VA, USA The items were not conditioned before use in order to simulate real-life usage 2.3 Sample generation and environmentally controlled room All of the indoor air samples were collected in the environmentally controlled room located at the Philip Morris International Research and Development facilities in Neuchâtel, Switzerland (Fig S2) This room has been described in detail in previous publications [31] All of the samples (except the background sample) were generated by means of three single-channel, programmable, dual-syringe pumps (PDSP, Burghart, Wedel, Germany) The TSNA validations were undertaken using a simulation of “Residential” environmental conditions (category I adapted from the EN standard 15251:2007) [40], characterized by a ventilation of 121 m3 /h corresponding to 1.67 air changes per hour Two fans were used to mix and distribute the air in the room The humidity was monitored, and the temperature was set to 23 ◦ C ± ◦ C The environmental aerosols of THS 2.2 and the ETS samples were generated under the Health Canada Intense machine-smoking regime with 12 and 10 puffs for the THS 2.2 tobacco stick and cigarette, respectively () [41] Three test items were used per hour, for a total of 12 test items used over the four hours of sample trapping The environmental aerosol of the e-cigarette samples was generated under the CORESTA machine-smoking regime () [42] One test item (50 puffs) was used per hour, for a total of four test items used over the four 92 M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 hours of sample trapping During the ETS generation sessions, the cigarette sidestream smoke was delivered to the environmentally controlled room while the mainstream smoke was drawn out of the room (surrogate ETS) For THS 2.2 and e-cigarettes, the entire mainstream aerosol was delivered to the environmentally controlled room (surrogate environmental aerosol) 2.4 Internal standards and standards (calibration and spiking) preparation The internal standard solution was prepared by adding 100 ␮L of NNN-D4 and NNK-D4 commercial solutions to a 1000 mL volumetric flask containing 990 mL of MeOH and diluting the mixture to the volumetric flask volume The NNN and NNK stock solutions were produced independently by transferring 800 ␮L of the certified reference material to 25 mL volumetric flasks containing 20 mL of MeOH and diluting the mixtures to the volumetric flask volume Standard level was prepared by transferring 190 ␮L of NNK and NNN stock solution into a 100 mL volumetric flask containing 90 mL of internal standard solution and diluting the mixture to the volumetric flask volume The calibration standards from level to level 1, as well as the spiking standards, were prepared by dilution of the higher-concentration standard solutions The typical concentration for the NNN calibration standards ranged from 0.196 ng/mL of level to 60.2 ng/mL of level 8, and for the NNK calibration, the standards ranged from 0.197 ng/mL of standard to 60.6 ng/mL of standard (Tables S1 and S2) Two sets of spiking standards were prepared, each one intended for the expected concentration of the target compounds to be measured in the different matrices The spiking concentrations were determined during the prevalidation phase, based on the type of matrix to be evaluated The first set of spiking solutions was in the range of 0.23 ng/mL to ng/mL for both NNN and NNK This set was used for the background and environmental aerosols of THS 2.2 and e-cigarettes A second set of spiking solutions was prepared for ETS (0.99 ng/mL to 30.3 ng/mL) 100 ␮L of each spiking solution was added to each cartridge, containing one of the four matrices of interest (Table S3) 2.5 Determination of TSNA 2.5.1 Description of the analytical method The procedure for running the method in routine is as follows: NNN and NNK are collected for two to four hours at a nominal flow-rate of 1.5 L/min on Sep-Pak Silica 690 mg sorbent cartridges (Waters Corporation) Post-collection, an amount of 100 ␮L of the NNN-D4 and NNK-D4 internal standard solution is added to the cartridges The TSNAs are eluted from the cartridges through a 0.2 ␮m polyvinylidene fluoride filter (Millipore Corporation) with 3.5 mL ethyl acetate and collected in mL cryogenic vials (Corning Inc.) The cryogenic vials containing the TSNA solution are placed on a thermal concentrator (Stuart), and the solvent is evaporated to dryness under a nitrogen flush over a period of approximately 35 The residuals are dissolved by addition of 100 ␮L of methanol to the cryogenic vial that, once capped, is vortexed briefly The obtained solutions are transferred into inserts, placed in amber LC vials, and then capped (Fig S3) Two ␮L of the solutions are injected and analyzed by liquid chromatography coupled with a triple quadrupole mass spectrometer (LC-ESI–MS/MS 5500 QQQ, ABSciex, Framingham, Massachusetts, USA) equipped with a heated nebulized interface in positive ionization mode A gradient separation is performed on a Kinetex pentafluorophenyl propyl (PFP) column (50 × 2.1 mm, 1.9 ␮m) HPLC column (Phenomenex), using 0.1% formic acid in water as mobile phase A and mL of formic acid into 90% methanol LC–MS grade/10% tetrahydrofuran (THF) as mobile phase B The details are presented in Table S4 The analytes are detected by multiple-reaction-monitoring using compound-dependent parameters (Table S5) The source temperature is set at 600 ◦ C, the ion source gas is set at 30 [AU], the nebulizer current is set at 5500 V, the collision gas is set at [AU], and target scan time is set at 0.27 s (Table S6) The method is accredited under ISO 17025 by the Swiss Accreditation Service (SAS) (STS 0045, SAS, Bern, Switzerland) 2.5.2 Validation design The validation was designed to assess all the method performance parameters as a function of the matrices The evaluated parameters were selectivity, linearity, and integrity of the response function; instrumental limit of detection (LOD), lower limit of quantification (LLOQ), and upper limit of quantification (ULOQ); repeatability limit and instrumental repeatability; intermediate precision (IP) limit; critical difference (CD); recovery; working range; and uncertainty To evaluate the matrix effect on the performance of the method, the use of spiked samples was selected The validation data were acquired by using unspiked homogenized and non-homogenized samples as well as spiked homogenized samples All of the samples were collected over a period of four hours and extracted, as described in 2.5.1 Four cartridges were used for the preparation of two different types of solutions: homogenized or non-homogenized samples For the homogenized samples, the eluents from four different cartridges were collected in a larger container and mixed well; the solution was subsequently split among four different cryogenic vials (Fig S3) The extract of each cartridge was collected individually for the non-homogenized samples To prepare the samples with internal standard (homogenized and non-homogenized), an amount of 100 ␮L of the NNN-D4 and NNK-D4 internal standard solution was added to the cartridges For the samples without internal standard (homogenized and nonhomogenized), 100 ␮L of MeOH were added instead The spiked samples were produced by spiking 100 ␮L of the different spiking standards (containing internal standard) directly on the cartridges (see section 2.4) Four series for each type of sample were collected per compound and matrix To be able to obtain sufficient measurement solutions to prepare all necessary samples, two days of sampling were required to complete one series Each day, 26 test portions were collected on the cartridges On the first day of each series, the samples collected were used to produce the homogenized matrix, non-homogenized matrix, and spiking levels 1–3 On the second day of the series, the samples collected were used to prepare the homogenized matrix and spiking levels 4–6 Four replicates were analyzed per sample type or level of spike The spiking ranges were adapted according to the samples’ endogenous content of each TSNA, as they differed substantially in the four different matrices (Table S7) The sample preparation and the analysis were performed by two trained operators A summary of all formulas used for the statistical computations is given in Table S8 Results and discussion The following section is divided into two parts The first is related to the method development phase, and the second is focused on the validation of the TSNA quantification method 3.1 Method development 3.1.1 Analytical method The development was initiated based on an internal method for mainstream aerosol analysis and several publications [13,15–18] M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 The instrument chosen for the analysis was an LC-20A Prominence Shimadzu HPLC system coupled with an ABSciex LCESI–MS/MS 5500 QQQ Electrospray ionization mode provided increased sensitivity for NNN and NNK compared with the use of atmospheric pressure chemical ionization For the chromatographic separation, a C18 reverse phase column, as described previously [13,16–18], was initially evaluated Although the chromatographic separation was relatively satisfactory, and the baseline separation of NNN and NNK with ␣ = 1.07 was achieved (Table S9 and Fig S4), additional columns were tested to further improve the peak separation and reduce the total run time As previously reported, columns with a PFP stationary phase give promising results in the analysis of TSNAs [15] The initial choice was to select the Pinnacle DB PFP (50 mm x 2.1 mm, 1.9 ␮m) column with base-deactivated spherical silica, as it demonstrated good retention for amine-containing compounds as well as efficiency with acidic mobile phases and highly aqueous mobile phases Indeed, the test showed an improvement in the peak separation (␣ = 1.64) and a significant reduction in the total run time (Table S10 and Fig S5) However, further tests demonstrated that the column deteriorated quickly, with a decrease in analytes retention After evaluation of several columns with the same type of phase (Nucleodur PFP 50 mm × 2.0 mm, 1.8 ␮m; Discovery HS F5 50 mm × 2.1 mm, ␮m; Luna PFP (2) 50 mm × 2.0 mm, ␮m; Kinetex PFP 50 mm × 2.1 mm, 1.7 ␮m; Express F5 50 mm × 2.1 mm, 2.7 ␮m), a Kinetex PFP 100A 50 mm × 2.1 mm, 1.9 ␮m column was finally selected As mobile phase, water/methanol solutions with mM ammonium acetate or using formic acid as modifier were tested, and isocratic and gradient conditions were evaluated Using acidified mobile phases A and B (0.1% formic acid in water as mobile phase A and 0.1% of formic acid in methanol as mobile phase B) led to a substantial increase in the peak intensities of NNN and NNK The addition of THF to mobile phase B increased the resolution between the NNN and NNK peaks The best chromatographic performance in terms of peak resolution (between 2.66 and 5.66), peak symmetry (peak width at 50% height between 0.0477 and 0.0915 for NNN and between 0.0782 and 0.102 for NNK), and duration of the analytical run (11 with column purge and equilibration) was achieved with the parameters given in Table S4 A typical chromatogram is presented in Fig 3.1.2 Aerosol collection and sample preparation The commonly used process of trapping on glass fiber Cambridge filter pads was the initial approach selected for trapping Tests based on the aerosol collection process and the sample preparation published by Tricker et al [13] and Wu et al [16] were performed This procedure included conditioning of the Cambridge filter pads, treatment with ascorbic acid, and irradiation with UV light After sampling, a triple extraction of each Cambridge filter pad was conducted, followed by combination of the extracts, evaporation, and reconstitution After dissolving the residue, liquid–liquid extraction with neutralization of the water phase was performed, followed by solid-phase extraction with evaporation of the solvent, and again reconstitution The main disadvantage of this approach was the laborious sample preparation, which adversely affected sample throughput In addition, as the peak intensities of NNN and NNK in the matrix samples were very low, some recovery inefficiencies were suspected To verify this, the Cambridge filter pads were spiked with the calibration solutions (containing the NNN and NNK standards as well as the deuterated compounds) to evaluate the recovery Very poor yields (under 37%) were obtained (Table S11) Different tests were subsequently carried out to improve and shorten the process Nevertheless, even with improved recoveries (75–80%), the sample preparation remained time-consuming 93 Fig Typical chromatogram for NNN and NNK The blue trace represents the NNN transition used as quantifier (178/148) The red trace represents the NNN transition used as qualifier (178/120) The green trace represents the NNN-D4 transition (182/152) The grey trace represents the NNK transition used as quantifier (208.1/121.7) The light blue trace represents the NNK transition used as qualifier (208.1/79) The pink trace represents the NNK-D4 transition (211.8/126) To resolve this issue, a novel approach for TSNA sample collection was evaluated The following considerations were taken into account TSNAs are present in the particulate phase of mainstream aerosol [11] and suspected to be distributed between the particulate phase and gas phase of environmental aerosols [11,43] Considering that TSNAs are polar compounds with an affinity for polar sampling media, silica traps were evaluated Sep-Pak Silica Classic Cartridges (690 mg) from Waters were tested The main advantages of using this alternative trapping were the possible reduction in the solvent volume used for the extraction, the possibility of removing the solid phase extraction process, and the overall simplification of the sample preparation procedure The first comparative test included trapping of ETS on Cambridge filter pads and on silica cartridges, both at L/min for four hours The results were encouraging and indicated yields in ETS in the same order of magnitude for both NNN and NNK (Table S12) To optimize this new trapping procedure, further investigations were performed Different collection flow-rates and times were considered to evaluate breakthrough and define the best trapping conditions No breakthrough occurred with trapping for four hours at 1.5 L/min, and the amount of constituents increased proportionally with respect to the values obtained for two hours of trapping Several extraction solvents were evaluated (Table S13) Ethyl acetate was selected due to its high volatility and the improved recovery compared with the other solvents (Table S13) In addition, 120 mg Sep-Pak Silica Classic Cartridges from Waters were also tested They were discarded due to back-pressure issues at a flow-rate of 1.5 L/min (Table S14 and Fig S6) Moreover, higher recoveries were achieved by filtration during elution of the sampling cartridge instead of the reconstituted solution before conducting instrumental analysis 3.2 Method validation 3.2.1 Selectivity To assess the selectivity of the method for the internal standard (NNN-D4 and NNK-D4 ), a comparison was made of the chromatograms of different blank samples (solvent and cartridges), calibration standards, and indoor air samples (background environ- 94 M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 mental aerosols of THS 2.2 and e-cigarettes, and the ETS of Marlboro Gold) with and without internal standard For the evaluation of the method selectivity for the target compounds, solvent and cartridge blank chromatograms were compared with those of standard To identify unequivocally the chromatographic signals corresponding to NNN and NNK in the matrices, spiking experiments were carried out to assess the increase in the signal In all of the solutions without internal standard, the possible interference peak area signals were below 10% of the internal standard peak area (Figs S7–S11) Furthermore, the measurement of blank samples (e.g., solvents) and blank collection trap signals in the area of the target compounds never exceeded the first calibration standard concentration (Figs S7 and S8) A linear increase in the signals for NNN and NNK in the chromatograms of the matrices was observed when adding spiking solutions at different concentrations (Figs S12 and S13), and the deviation in the retention time of the peaks for NNN and NNK, as well as the NNN-D4 and NNK-D4 through the sequences, was within the established acceptance range of ±0.20 Thus, selectivity proved to be satisfactory for all target compounds and internal standards 3.2.2 Linearity and integrity of the response function For both compounds, the response function was determined by an examination of all calibration curves injected during the validation (one calibration curve per sequence, 32 in total) The standards used for the calibration curves contained both internal standards (NNN-D4 and NNK-D4 ) A weighted (1/concentration), not forcedthrough-origin, quadratic response of the type y = ax2 + bx + c was the most suitable to describe the relationship between the measured concentration (x) and the area ratio between each target compound and the respective internal standard (y) based on the results obtained for the residuals (Table S15; Figs S14 and S15) All the calculated determination coefficients were above 0.9989, and the residuals, per level of concentration, measured during the whole validation never exceeded ± 20% (Table S16; Figs S16–S23) To evaluate the integrity of the response function, a vial of standard level was injected several times through each sequence, and the results remained within ± 20% of the theoretical concentrations for all the sequences 3.2.3 Instrumental LOD, LLOQ, and ULOQ The values of the LLOQ were calculated by multiplying the standard deviation of all the concentrations measured for standard during the validation by 10 (0.129 ng/mL for NNN and 0.162 ng/mL for NNK) [44] The results for both compounds were lower than their respective standard concentrations; therefore, standard values were defined as the method LLOQ (0.196 ng/mL/0.0544 ng/m3 for NNN and 0.197 ng/mL/0.0547 ng/m3 for NNK) The values of the LOD were calculated by dividing the calculated LLOQ by 3.3 (0.0390 ng/mL/0.0108 ng/m3 for NNN and 0.0491 ng/mL/0.0136 ng/m3 for NNK) The ULOQ was set as the highest calibration standard tested for which the respective calibration fulfils the acceptance criteria related to the linearity of the response function [44] All level working standards analyzed during the validation fulfilled the acceptance criteria set for the response function; therefore, the concentration of standard was considered as method ULOQ (60.2 ng/mL/16.7 ng/m3 for NNN and 60.6 ng/mL/16.8 ng/m3 for NNK) (Table S17) The method described here proved superior in terms of sensitivity as illustrated by comparison with the LOD and LLOQ of published methods (Table S18 in Supporting Information) Indeed, both the LOD and LLOQ were one to two orders of magnitude below those reported for the analyses of NNN (LOD: 0.625 ng/m3 ; LLOQ: 2.06 ng/m3 ) and NNK (LOD: 0.750 ng/m3 ; LLOQ: 2.48 ng/m3 ) in air (see details in Table S18 in the Supporting Information) [13] 3.2.4 Instrumental repeatability and repeatability limit r The measurement of the instrumental repeatability was performed by injecting the same calibration standard level (standard 3) six times through all of the sequences run for the analysis of every single matrix The coefficient of variation (CV) obtained for standard was compared with the value set forth in the FDA guidelines [44] For the assessment of the repeatability limit of the whole process, the CV obtained for the four non-homogenized samples of the different matrices per day of analysis was also compared with the reference values set forth in the FDA guidelines [44] The maximum repeatability coefficients of variation (withinday coefficient of variation, CV) measured per day for NNN and NNK were 5.8% and 7.4%, respectively, which fulfilled the 22% CVr set as acceptance criterion for the validation [44] The CVr (per day of analysis) was also measured for the matrix samples This parameter could not be evaluated for either background samples or NNK in the environmental aerosol of e-cigarettes CVr values for NNN in the environmental aerosol of THS 2.2, the environmental aerosol of e-cigarettes, and ETS were 6%, 18%, and 5%, respectively For NNK, the CVr was 8% for the environmental aerosol of THS 2.2 and 5% for ETS For both compounds, CVr never exceeded the maximum of 22% set as acceptance criterion [44] The repeatability limit (r) was determined by analysis of four matrix samples per day of analysis This parameter could not be assessed for background samples The r values for NNN in the environmental aerosol of THS 2.2, the environmental aerosol of e-cigarettes, and ETS were 0.0511 ng/mL, 0.102 ng/mL, and 0.397 ng/mL, respectively For NNK, the r values were 0.0459 ng/mL for the environmental aerosol of THS 2.2, and 1.96 ng/mL for ETS (Table S21) 3.2.5 Working range 3.2.5.1 Trueness Accuracy is the sum of two parameters: precision (determined by the intermediate precision, IP) and trueness (closeness between measured and reference values) [46] As no reference materials were available, evaluation of these parameters was performed by spiking experiments The cartridges containing the aerosol collection replicates were spiked with a known concentration of NNN and NNK and then extracted and analyzed The concentrations of the solutions used for spiking were aligned according to the Association Franc¸aise de Normalisation (AFNOR) norm [45] when the quantities for producing the spiking solutions varied from one day to another The endogenous content of NNN and NNK already measured in the matrices (if any) was subtracted from the content measured for the spiked samples to calculate the recoveries As this endogenous content measured for the matrices varied between spiking levels (spike levels 1–3 were analyzed on day 1, levels 4–6 were analyzed on day 2) and from one series to another, the endogenous content of the non-spiked homogenized matrix collected the same day was subtracted for the recoveries calculation The average endogenous content of NNN and NNK in the background matrix was below LOD NNN average measured values for the environmental aerosol of THS 2.2, the environmental aerosol of e-cigarettes, and ETS were 0.306 ng/mL, 0.201 ng/mL, and 2.95 ng/mL, respectively For NNK, the average endogenous content values were 0.246 ng/mL for the environmental aerosol of THS 2.2 and 14.8 ng/mL for ETS The average recoveries in the background matrix were between 102% and 131% for NNN and between 106% and 125% for NNK In the environmental aerosol of THS 2.2, the average recoveries for both compounds were between 96% and 99% NNN average recoveries M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 95 Table Data used to build the accuracy profiles for NNN Matrix1 BKG EA of THS2.2 ETS EA fo e-cig Spiking level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Spiking concentration [␮g/mL] 0.229 0.514 0.883 1.66 3.41 4.97 0.229 0.514 0.883 1.66 3.41 4.97 0.989 1.977 4.945 7.97 15.93 30.10 0.229 0.514 0.883 1.66 3.41 4.97 Trueness values per series Average 142 107 117 109 103 103 131 101 113 103 103 102 89 91 98 104 106 100 94 96 96 106 103 104 107 101 95 109 104 101 30 64 70 93 94 94 119 101 95 97 99 99 123 118 110 109 103 105 144 126 114 106 103 103 135 114 113 95 96 96 103 106 99 103 98 98 93 103 101 105 101 102 131 113 107 105 100 100 90 112 100 93 95 96 130 93 90 94 99 97 72 96 100 105 104 105 131 112 108 107 103 102 96 98 99 96 97 97 110 97 95 99 101 98 95 103 102 106 103 104 CVr CVR 2 14 2 2 16 10 3 18 11 10 2 50 24 20 5 20 25 14 3 80% Tolerance Interval Lower limit Upper limit 100 93 90 104 99 99 56 62 87 90 91 78 84 88 91 94 95 56 83 90 102 99 101 161 131 126 111 106 105 186 139 136 105 105 103 143 110 103 108 107 102 135 124 113 111 107 107 BKG: Background, EA: Environmental Aerosol, ETS: Environmental Tobacco Smoke in the environmental aerosol of e-cigarettes were in the range of 95%–106% for NNN and in the range of 99%–114% for NNK The average recoveries measured in ETS were between 95% and 110% for NNN and between 96% and 130% for NNK (Tables S22 and S23) 3.2.5.2 Intermediate precision limit (IP) and critical difference (CD) The evaluation of the repeatability and IP was performed by analyzing the matrix samples collected on four different days According to the FDA guidelines [44] and AFNOR NF V 03-110 [45], the acceptance criterion was set as concentration-dependent Therefore, for concentrations lower than 10 ppb, no limit was set, and for concentrations in the range of 10 ppb, an initial limit of ±35% was set [44,45] The measured concentrations of the non-homogenized matrices injected on different days were compared with the values set forth the in the FDA guidelines [44] and AFNOR NF V 03-110 [45] per level of concentration The critical difference was calculated based on these values If the concentrations of NNN and NNK were below LOD or standard 1, CD values could not be calculated (NNN and NNK in the background matrix, NNK in the environmental aerosol of ecigarettes) The CD values for NNN in the environmental aerosol of THS 2.2, the environmental aerosol of e-cigarettes, and ETS were 0.162 ng/mL/0.0449 ng/m3 , 0.268 ng/mL/0.0744 ng/m3 , and 1.23 ng/mL/0.343 ng/m3 , respectively For NNK, the CD values were 0.135 ng/mL/0.0374 ng/m3 in the environmental aerosol of THS 2.2 and 14.1 ng/mL/3.92 ng/m3 in ETS (Table S24) 3.2.5.3 Accuracy profiles Validation of the TSNA method was performed using the accuracy profile procedure [44–46] This validation procedure was considered as the most appropriate to evaluate the analytical method performance in each matrix under investigation (background, environmental aerosol of THS 2.2 and e-cigarettes, and ETS) per target compound The trueness (recovery) per level of spike was calculated and, together with the intermediate precision and the tolerance interval, was used to build the accuracy profiles per compound and matrix [44–46] The lower working range limit (LWRL) and upper working range limit (UWRL) were calculated after evaluation of the obtained accuracy profiles For the validation of NNN and NNK in the four matrices, the ␤-expectation tolerance intervals and the acceptance limits were set at 80% and ± 25%, respectively One graph was generated for each target compound and matrix combining the corresponding tolerance interval and acceptance limit On every graph, the ± 25% acceptance limits are represented by horizontal, red, dotted lines The trueness is represented by an orange, small-striped line connecting the average percentage of recovered concentration per spiking level, depicted by dots The uncertainty per spike level is presented by interval (black vertical lines), and the two solid blue lines at both sides of the trueness are the representation of the 80% ␤-expectation tolerance limits The vertical, green-striped, dotted line indicates the cut point between the 80% ␤-expectation tolerance limits and the ± 25% acceptance limits This cutting point corresponds to the LWRL per compound and matrix The average measured concentrations per matrix type (endogenous amount) are represented by green, square dots Fig presents the NNN accuracy profiles per matrix type and Table contains the data used to build the accuracy profiles Fig presents the NNK accuracy profiles per matrix type and Table contains the data used to build the accuracy profiles The impact of the matrices (background, environmental aerosols of THS 2.2 and e-cigarettes, and ETS) on the performance of the methods was negligible, as similar performances (e.g., comparable ␤-tolerance intervals and a lack of bias) between all matrices were observed 3.2.5.4 Working range limits The LWRL and UWRL for the two target compounds in each matrix were determined using the accuracy profiles In all cases, the LWRL was defined by the intersection point between the ␤-tolerance interval and the acceptance limits [46] At higher concentrations (spiking levels 4–6), the ␤-tolerance interval remained between the acceptance limits for all four matrices Therefore, the UWRL was defined as the highest calibration level fulfilling the criteria for linearity of the response function (60.2 ng/mL for NNN and 60.6 ng/mL for NNK, see section 3.2.3) 96 M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 Table Data used to build the accuracy profiles for NNK Matrix1 BKG EA of THS2.2 ETS EA fo e-cig Spiking level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Level Spiking concentration [␮g/mL] Trueness values per series 0.230 0.517 0.889 1.67 3.43 5.00 0.230 0.517 0.889 1.67 3.43 5.00 0.995 1.989 4.975 8.01 16.03 30.28 0.230 0.517 0.889 1.67 3.43 5.00 Average 127 107 119 114 110 108 110 86 106 105 104 104 131 66 106 95 105 98 122 123 113 104 96 102 89 92 90 110 105 105 87 94 91 101 98 97 169 129 103 104 107 102 129 121 111 103 99 104 163 142 126 112 109 106 110 121 110 96 98 101 59 127 79 95 90 92 112 87 93 115 107 105 121 104 100 104 101 103 76 92 87 92 94 97 163 166 113 92 99 99 92 104 109 95 93 95 125 111 109 110 106 106 96 98 99 98 99 99 130 122 100 96 101 98 114 109 107 104 99 101 CVr CVR 7 2 11 42 13 21 17 12 3 31 22 18 20 16 12 62 43 16 25 22 14 80% Tolerance Interval Lower limit Upper limit 70 72 78 102 99 102 64 70 78 88 91 93 34 47 73 86 86 90 79 76 87 90 88 93 180 150 139 118 113 110 127 126 119 109 106 106 226 198 127 107 115 106 149 141 127 119 110 109 BKG: Background, EA: Environmental Aerosol, ETS: Environmental Tobacco Smoke Table Lower Working Range Limits (LWRL) and Upper Working Range Limits (UWRL) for NNN and NNK Matrix1 BKG EA of THS2.2 ETS EA of e-cig Min Max Target compound NNN NNK NNN NNK NNN NNK NNN NNK NNN NNK LWRL2 UWRL2 ng/mL ng/m ng/mL ng/m3 0.919 1.30 1.16 0.702 1.37 5.24 0.453 0.994 0.453 0.702 0.255 0.362 0.322 0.195 0.379 1.46 0.126 0.276 0.126 0.195 60.2 60.6 60.2 60.6 60.2 60.6 60.2 60.6 60.2 60.6 16.7 16.8 16.7 16.8 16.7 16.8 16.7 16.8 16.7 16.8 BKG: Background, EA: Environmental Aerosol, ETS: Environmental Tobacco Smoke Conversion from ng/mL to ng/m3 using 1.5 L/min sampling flow-rate and four hours of collection (0.36 m3 ), and final solution volume of 0.1 mL The LWRL for NNN in the background matrix, the environmental aerosol of THS 2.2, the environmental aerosol of e-cigarettes, and ETS were 0.919 ng/mL, 1.16 ng/mL, 0.453 ng/mL, and 1.37 ng/mL, respectively For NNK, LWRL was 1.30 ng/mL in background, 0.702 ng/mL in the environmental aerosol of THS 2.2, 0.994 ng/mL in the environmental aerosol of e-cigarettes, and 5.24 ng/mL in ETS Table presents the LWRL and UWRL for the two target compounds in the four matrices under investigation 3.3 Application of the quantification method NNN and NNK have attracted considerable research interest due to their demonstrated carcinogenicity in animal models and their assumed contribution to the overall carcinogenic potential of tobacco smoke [2,46–48] The World Health Organization study group on Tobacco Product Regulation has identified NNN and NNK as two of the nine priority smoke components of regulatory interest [49] In view of this, the use of a sensitive and accurate method for the measurement of these compounds is of key importance NNN and NNK are emitted at the same concentrations as in mainstream smoke or even at two to four times higher concentrations in cigarette sidestream smoke, which is the predominant component of ETS [11] According to published data, 84%–97% of NNN and 63%–84% of NNK present in mainstream smoke of cigarettes are retained in the lungs of the consumers [50] The environmental aerosols of heat-not-burn products and ecigarettes have different origin and characteristics compared with ETS, because by design, these products not have a smoldering tip releasing sidestream smoke resulting from combustion of organic material The main component of their environmental aerosols is thus the exhaled breath of the users Accordingly, considering the high retention in the body of these compounds from mainstream smoke, it is anticipated that air concentrations of NNN and NNK will be very low following the use of heat-not-burn products and e-cigarettes This newly developed and validated method was put into use for the quantification of NNN and NNK in indoor air enriched with surrogate environmental aerosols generated by smoking machines Thus, aged and diluted mainstream aerosols of heat-not-burn products and e-cigarettes were released in the environmentally controlled exposure room, while for cigarettes, aged and diluted sidestream smoke was released In such an experimental setup, the environmental impact of heat-not-burn products and e-cigarettes is overestimated In ETS samples, NNK and NNN were quantified during all the sessions with values between the LWRL and the UWRL (NNN: 0.816 ng/m3 , NNK: 4.13 ng/m3 ) (Tables 4, S25) Table presents the average NNN and NNK matrix endogenous content for homogenized and non-homogenized samples of the four matrices under investigation The concentrations of NNN and NNK in indoor air during cigarette smoking were investigated in experimental rooms [12,13,22,23,34] as well as in real-life conditions [12] In experiments with cigarette smokers, the quantified indoor levels for NNN ranged from not detected to 23 ng/m3 , and those for NNN ranged from not detected up to 29 ng/m3 ; however, most of the values measured were below 10 ng/m3 [12,13,22,23,34] For example, in M Gómez Lueso et al / J Chromatogr A 1580 (2018) 90–99 97 Table Average content of NNN and NNK in Background, environmental aerosol of THS 2.2 and e-cigarette, and environmental tobacco smoke (smoking machine model) Matrix1 BKG EA of THS2.2 EA of e-cig ETS Average STDEV Average STDEV Average STDEV Average STDEV Average endogenous content in homogenized matrix Average endogenous content in non- homogenized matrix Average endogenous content per matrix type (all values) NNN ng/m3 NNK ng/m3 NNN ng/m3 NNK ng/m3 NNN ng/m3 NNK ng/m3