Variations of TSNA Levels in Tobaccos Upon Heating at Moderate Temperatures

Several chemicals including ammonium acetate, ammonium formate, formic acid, and acetonitrile were obtained from Sigma/Aldrich (St. Louis, MO, USA). Nicotine, nitrosoanabasine (NAB), nitrosoanatabine (NAT), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), nitrosonornicotine (NNN), NAB-d₄, NAT-d4, NNK-d₄ and NNN-d₄ were obtained from Toronto Research Chemicals Inc. (TRC) (North York, ON, Canada). Pure water (18.2 MΩ/cm) was obtained from a Barnstead water purification unit (Thermo Fisher Scientific, Waltham, MA, USA). The GC vials were 2 mL with screw top caps with septa. For filtration 0.45-μm PVDF filters were used (Whatman Autovial, GE Healthcare, Little Chalfort, UK).

The instruments used for the analysis consisted of an Agilent 1290 HPLC binary system with a binary pump, an autosampler with cooling capability, and a column thermostatted compartment. The HPLC chromatographic separation was achieved on a Luna® 3 μm C18 150 × 3 mm from Phenomenex (Torrance, CA, USA). The MS/MS system was an API-6500 triple quadrupole mass spectrometer (AB Sciex, Framingham, MA, USA). The LCMS/MS system was controlled using Analyst 1.6.2 software, and the peak integration was performed with MultiQuant 3.0.1 software. The heating was performed in a Thermoline Furnace 62700 (Thermo Fisher, Waltham, MA, USA). Tobacco moisture was measured with a halogen moisture analyzer HE53 (Mettler Toledo, Greifensee, Switzerland).

An extraction solution of 100 mM ammonium acetate in water was prepared by dissolving 7.7 g of ammonium acetate into 1 L of purified water. The standard solutions were prepared from a stock solution of 162.65 μg/mL N-nitrosonornicotine (NNN), 164.036 μg/mL N’-nitrosoanatabine (NAT), 41.508 μg/mL N-nitrosoanabasine (NAB) and 161.10 μg/mL nicotine-derived nitrosamine ketone (NNK). Standard 1 was prepared by 1/1000 dilution of the stock solution with the extraction solution and used to prepare Standards 2–7, 1000 μmL each. To each 1 mL standard was added 20 μL of the internal standard solution. The internal standard solution contained 1.6 μg/mL NNN-d₄, 1.6 μg/mL NAT-d₄, 1.33 μg/mL NAB-d₄ and 1.6 μg/mL NNK-d₄. The concentration of internal standard (I.S.) in the final solution is indicated in Table 1.

The concentrations of the utilized standards are given in Table 2.

Heating of the tobacco was performed in sealed Pyrex glass tubes. This was done to avoid potential loss of material by evaporation when the tobacco was heated. Glass tube dimensions were 8 mm i.d. and a length of 15 cm with a 1.0-mm wall thickness. About 600 mg tobacco (± 20 mg) were weighed into each tube. The tubes were sealed at one end and were subjected to a mild vacuum prior to sealing the other end. The purpose of applying a vacuum was to prevent the tubes from breaking when exposed to heat due to the expansion of gases in the tube. Figure 1 shows several sealed tubes containing powdered tobacco (Burley lower stalk) before heating.

Figure 1

Four different temperatures were used for heating the tubes: 100 °C, 150 °C, 200 °C, and 250 °C. Two time intervals were selected for heating: 2 min and 5 min. The actual temperature of the tobacco was not measured during the experiments performed for TSNAs measurement. However, in a published study (3) the heating inside of a tobacco rod was evaluated, when exposed from exterior to 100 °C or 200 °C. The study (3) indicated a delay in the temperature reached by the middle of a tobacco rod as compared to the applied external temperature. In the present study, an additional issue was that upon removing the glass tube with the tobacco from the oven, the cooling of the material was not instantaneous and it took 3–4 min for the tube to reach room temperature. For this reason, the temperature inside a sample (of Burley) was measured separately as a function of heating time. For this purpose a thermocouple was inserted into the sample during the heating of the glass tube containing the sample. At 5 min the heating was stopped by removing the glass tube from the oven and exposing it to room temperature. The temperature variation inside the sample is shown in Figure 2.

Figure 2

As indicated in Figure 2, the temperature to which the glass tube with sample is exposed is different from the actual temperature of the sample. Assuming a linear idealized temperature profile to which the tobacco is exposed, the temperature variation in the tobacco can be described by the following formula: [1] T(t)=Tinit.+b1 t(0<t<tmax)T(t)=Tinit.+b1 tmax−b2 t(tmax<t<tfinal) \begin{array}{*{20}{c}} {T\left( t \right) = {T_{{\text{init}}.}} + {{\text{b}}_1}\,\,t\left( {0 < t < {t_{{\text{max}}}}} \right)} \\ {T\left( t \right) = {T_{{\text{init}}.}} + {{\text{b}}_1}\,\,\,\,{t_{{\text{max}}}} - {{\text{b}}_2}\,\,\,\,t\left( {{t_{{\text{max}}}} < t < {t_{{\text{final}}}}} \right)} \end{array} where T(t) is the temperature of the sample as a function of time t, T_init. is the initial temperature (about 22 °C), t_max is the length of time the sample is heated, t_final is the time of cooling to room temperature, and b₁ and b₂ are heating and cooling rates, respectively. The fixed oven temperatures 100 °C, 150 °C, 200 °C, and 250 °C for which the results are indicated in this study, are in fact described correctly by sample temperature variations as shown in Figure 2 and approximated by equation [1]. However, the linear approximation holds only for the first minute of heating. Afterwards the higher temperatures showed an endothermic leveling when passing 100 °C and the rates for all temperatures decreased as the difference in temperature between the tobacco and the outside of the tube decreased. The initial slopes (°C/s) were proportional to the oven temperature and were equal to [0.0076 × oven temperature – 0.494] with an R² = 0.999. The initial rate of cooling was also linear and equal to [− 0.0062 × oven temperature + 0.433] with an R² = 0.939. The cooling rate was slightly lower than the heating rate because the initial temperature difference was lower between the inside and the outside of the tube. The fixed temperatures of the oven, to which the glass tubes were exposed, were precisely controlled (within 1–2 °C) while the true temperature of the samples was subject to variations depending on the exact mass of the glass tube, the mass of the tobacco sample, and on the distribution of the tobacco inside the glass tube during heating, all of which affected parameters b₁ and b₂. The use of sealed tubes as containers for the heated tobacco avoided potential loss of material by evaporation but at the same time did not allow a precise measurement of the temperature in the tobacco sample. This temperature can be inferred only by comparing the temperature of the heating of the glass tube with that inside the sample shown in the graph of Figure 2.

Prior to the analysis, the moisture of the samples was measured using a halogen moisture analyzer. For the analysis, samples of 250 ± 1.0 mg of tobacco (as is) were weighed directly into 20-mL scintillation vials. To each vial 10 mL of the extraction solution were added, then the vial was capped and placed into a shaker for 30 min. The solution was then filtered through a syringe filter into a test tube. 1000 μL of the filtered solution was transferred to a 2-mL GC vial and 20 μL of the TSNA internal standard was added to the sample.

The HPLC separation of TSNAs was performed using a gradient program. The mobile phase A was 100 mM ammonium formate pH 4.9 aqueous solution with 5% acetonitrile and mobile phase B was 0.2% formic acid in acetonitrile. The timetable for the gradient is shown in Table 3.

The parameters for the MS/MS detection included: curtain gas CUR = 20 mL/min, collision gas CAD = 4 mL/min, ion spray voltage IS = 4500 V, temperature TEM = 500 °C, ion source gas 1 GS1 = 40 mL/min, ion source gas 2 GS2 = 50 mL/min, entrance potential EP = 10 V, target scan time 0.11 s, scheduled MRM detection window 40 s. Other parameters including the analyzed ions are shown in Table 4.

Extracted ion chromatograms for the eight compounds measured in the study are shown in Figure 3 for a flue-cured tobacco (upper stalk) heated at 200 °C for 5 min. Both NNN and NNN-d4 show two peaks in the previously described separation conditions. Only the higher of the two peaks was used for the quantitation. Peak splitting for TSNAs can be attributed to the presence of a syn-anti isomers of the nitrosamine group.

Figure 3

For the quantitation of TSNAs, calibration curves were constructed by plotting the concentrations of the standard solutions as a function of the peak areas normalized to the area of the corresponding deuterated internal standards. All calibrations were linear and they were expressed using the formula: [2] Y(conc.)=a(Std. Area/I.S. Area)/b Y\left( {{\text{conc}}.} \right) = a\left( {{\text{Std}}.\,{\text{Area}}/{\text{I}.\text{S}}.\,{\text{Area}}} \right)/b

Parameters a and b as well as correlation coefficient R² for the calibration lines are shown in Table 5.

The overall validation of the analytical method for the analysis of TSNAs has been previously reported (18) and no further validation has been performed for the method.

The samples analyzed in this study are listed in Table 6, together with the measured moisture % total alkaloid level as well as the time intervals and temperatures to which they were exposed.

Some samples were processed in duplicate or even triplicate, and other samples were analyzed only once. The samples were randomly analyzed, and some duplicate or triplicate evaluations were made to verify the repeatability of the results. Once the sample was heated and extracted, the LC/MS/MS measurement was performed in duplicate.

The results of average TSNA levels in ng/g for the starting tobaccos (analyzed before starting the heating experiments) are shown in Table 7. All results are reported for the tobacco “as is,” a conversion to “dry basis” being possible using the moisture data from Table 6.

As indicated in Table 7, the levels of TSNAs did not differ meaningfully within tobacco type (by stalk position), but among the various tobacco types they differed considerably. Burley tobaccos have a very high level of TSNAs even before heating. Significantly lower levels compared to the Burleys are present in flue-cured tobaccos. The lowest TSNA levels are present in the Oriental tobaccos.

The experimental results for levels of TSNAs in the six evaluated tobaccos for the two heating intervals (2 min and 5 min) are indicated in Tables 8 to 10. In these tables, the listed and repeated temperatures (for a specific TSNA in each tobacco) indicate results from repeated whole experiments (including heating of the tobacco and LC/MS/MS measurement).

Each LC/MS/MS measurement for every experiment was performed in duplicate and the RSD% values are reported between repeated LC/MS/MS measurements.

The results of TSNA levels in heated flue-cured tobaccos FC-Lo and FC-Up are indicated in Table 8.

The results of TSNA levels in heated Burley tobaccos Bu-Lo and Bu-Up are indicated in Table 9. The results of TSNA levels in heated Oriental tobaccos are indicated in Table 10.