Quantitative Method for the Analysis of Tobacco-Specific

Jingcun Wu,*, Peter Joza,, Mehran Sharifi,, William S. Rickert, andJohn H. ... Matthew R. Holman , Liqin Zhang , Yan S. Ding , and Clifford H. Watson ...
16 downloads 0 Views 60KB Size
Anal. Chem. 2008, 80, 1341-1345

Quantitative Method for the Analysis of Tobacco-Specific Nitrosamines in Cigarette Tobacco and Mainstream Cigarette Smoke by Use of Isotope Dilution Liquid Chromatography Tandem Mass Spectrometry Jingcun Wu,*,† Peter Joza,† Mehran Sharifi,† William S. Rickert,† and John H. Lauterbach*,‡

Labstat International ULC, 262 Manitou Drive, Kitchener, Ontario, Canada N2C 1L3, and Lauterbach & Associates, LLC, 211 Old Club Court, Macon, Georgia 31210-4708

An improved liquid chromatography tandem mass spectrometry (LC-MS/MS) method has been developed for the determination of tobacco specific nitrosamines (TSNA). It utilizes four stable isotope-labeled internal standards instead of two as reported by others. A separate internal standard for each analyte is required to minimize sample matrix effects on each analyte, which can lead to poor analyte recoveries and decreases in method accuracy and precision if only one or two of the internal standards are used, especially for complex sample matrixes and when no sample cleanup steps are performed as in this study. In addition, two ion-transition pairs (instead of one) are used for each analyte for the confirmation and quantification, further enhancing the method’s accuracy and robustness. These improvements have led to a new LCMS/MS method that is faster, more sensitive, and selective than the traditional methods and more accurate and robust than the published LC-MS/MS methods. The linear range of the method was from 0.2 to 250 ng/mL with a limit of detection of each TSNA varied from 0.027 to 0.049 ng/mL. Good correlations between the results obtained by the new method and the traditional method were observed for the samples studied. Seven tobacco specific nitrosamines (TSNA) have been identified in both tobaccos (including cigarette filler) and cigarette smoke (see structures in Figure 1 of ref 1).1-4 Four of the most commonly analyzed TSNA compounds are N′-nitrosonornicotine * To whom correspondence should be addressed. E-mail, [email protected]; phone, 1-519-748-5409; fax, 1-519-748-1654 (J.W.). E-mail, john@ lauterbachandassociates.com; phone, 1-478-474-8818; fax, 1-478-474-0117. † Labstat International ULC. ‡ Lauterbach & Associates, LLC. (1) Wu, W.; Ashley, D. L.; Watson, C. H. Anal. Chem. 2003, 75, 4827-4832. (2) Wagner, K. A.; Finkel, N. H.; Fossett, J. E.; Gillman, I. G. Anal. Chem. 2005, 77, 1001-1006. (3) Lee, J.-M.; Shin, J.-W.; Oh, I.-H.; Lee, U.-C.; Rhee, M.-S. Determination of Tobacco Specific Nitrosamines in Mainstream Smoke of 2R4F Reference Cigarettes by LC/MS/MS. 2004 CORESTA Congress Kyoto, Kyoto, Japan, October 3-7, 2004; Paper SS20. Full text available on CORESTA CD-ROM Vol. 22; abstract available on the Internet at http://www.coresta.org/ Past_Abstracts/Kyoto2004-SmokeTech.pdf (accessed December 29, 2006). 10.1021/ac702100c CCC: $40.75 Published on Web 01/12/2008

© 2008 American Chemical Society

(NNN), N′-nitrosoanatabine (NAT), N′-nitrosoanabasine (NAB), and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK). Because of the health concerns and the progress made in reducing the TSNA content in tobacco products, it is necessary to develop a fast, sensitive, selective, and robust analytical method that can determine low levels of TSNA in both tobaccos and cigarette smoke. Traditionally, the most widely used method for TSNA analysis has been gas chromatography (GC) coupled with a thermal energy analyzer (TEA).3-5 However, the GC-TEA method has several disadvantages: (1) It cannot differentiate the coeluted nitroso compounds although it is nitroso-specific. (2) Extensive sample cleanup including liquid-liquid (L-L) extraction and solidphase extraction (SPE) are necessary in order to get good GC separation for the target compounds. (3) Samples must be concentrated before injection due to the limited sensitivity of the method. Recently, LC-MS/MS methods have been developed for TSNA analysis,1-4 providing greater sensitivity and selectivity over the GC-TEA method. However, one of the major concerns for the validation of a LC-MS/MS method is the sample matrix effects, which can lead to poor analyte recoveries and decreases in method’s accuracy and precision.6-10 One way to reduce these effects is to perform sample cleanup, such as L-L extraction and SPE, as shown in one of the LC-MS/MS methods for TSNA.1 However, this will increase the analysis time and limit the sample throughput. Another effective way to compensate for the matrix effects is to use stable isotope-labeled internal standards for calibration, which can maintain high-throughput analysis without (4) Chwojdak, C. A.; Self, D. A.; Wheeler, H. R. A Collaborative, Harmonized LC-MS/MS Method for the Determination of Tobacco Specific Nitrosamines (TSNA) in Tobacco and Tobacco Related Materials. 61st Tobacco Science Research Conference, Charlotte, NC,September 24, 2007. (5) Health Canada methods are available on the Internet at http://www.hcsc.gc.ca/hl-vs/tobac-tabac/legislation/reg/indust/method/index_e.html (accessed December 29, 2006). (6) Jessome, L. L.; Volmer, D. A. LCGC 2006, 24 (5), 498-510. (7) Guo, X.; Bruins, A. P.; Covey, T. R. Rapid Commun. Mass Spectrom. 2006, 20, 3145-3156. (8) Kebarle, P.; Tang, L. Anal. Chem. 1993, 65, 972A-984A. (9) McMaster, M. C. LC/MS: A Practical User’s Guide; John Wiley & Sons: Hoboken, NJ, 2005. (10) Food and Drug Administration. Guidance for Industry on Bioanalytical Method Validation. Fed. Regist. 2001, 66 (100), 28526 (Docke No. 98D1195).

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008 1341

sample cleanup. In addition to the sample matrix effects, the use of individual internal standards is also important to deal with solvent variations during gradient LC. However, the published LC-MS/MS methods1-3 use only two isotope-labeled compounds (NNN-d4 and NNK-d4) as internal standards for the four analytes, which can result in errors in the determinations of NAT and NAB due to the significant analyte-dependent matrix effects found in TSNA analysis, especially when applying these methods to various complex sample matrixes. To obtain accurate results across a wide range of sample matrixes, an improved LC-MS/MS method has been developed and validated in this study, using four isotopelabeled internal standards to match the four TSNA compounds determined. In addition, two ion transition pairs are used for each analyte in the new method for analyte confirmation and quantification, further enhancing the method’s accuracy and robustness.10,11 Sample preparations involve only analyte extraction with aqueous buffer followed by filtration using a syringe filter. These improvements allow us to take full advantages of the fast analysis, high sensitivity, and selectivity of LC-MS/MS method without need of extensive sample cleanup. The method can be used to determine the TSNA levels in a wide range of samples including both tobaccos and mainstream cigarette smokes. This new method together with the traditional GC-TEA method5 has been applied to TSNA analyses for Kentucky Reference cigarette (KY2R4F) and some brands of Canadian cigarettes collected in 2003, 2004, and 2005. EXPERIMENTAL SECTION Caution. The TSNA and related chemicals and samples should be handled according to the guidelines for Laboratory Use of Chemical Carcinogens.12 Reagents. TSNA standards and the four internal standards were obtained from Toronto Research Chemicals Inc., North York, ON, Canada, and used as received. These standards were dissolved in acetonitrile and diluted to make intermediate stocks (400 ng/mL for NNN, NAT, and NNK, and120 ng/mL for NAB), from which the calibration standards were prepared in the range of 0.8-200 ng/mL for NNN, NAT, and NNK and 0.2-60 ng/mL for NAB. All other chemicals were of analytical grade and obtained from Sigma. The Kentucky reference cigarette (KY2R4F) was obtained from the Kentucky Tobacco Research and Development Center (Lexington, KY). Several brands of commercially available Canadian flue-cured filter cigarettes from 2003, 2004, and 2005 were selected for this study. The four brands reported in this paper are described in Table 1 using the declared values from the packaging. All cigarettes were conditioned according to ISO 3402 prior to analyses.13 General Considerations for Sample Preparation and Analysis. The mainstream smoke collections (ISO and Canadian intense smoking conditions) from seven replicates for each brand (11) Official Journal of the European Communities L221, 8-36. Commission Decision (2002/657/EC) of 12 August 2002 Concerning the Performance of Analytical Methods and the Interpretation of Results. Brussels, Belgium, 2002. (12) NIH Guidelines for the Laboratory Use of Chemical Carcinogens; NIH Publication 81-2385; U.S. Government Printing Office: Washington, DC, 1981. (13) ISO 3402: 1999, Tobacco and Tobacco Products- Atmosphere for Conditioning and Testing, 4th ed.; International Organization for Standardization: Geneva, Switzerland; pp. 1-4.

1342 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

Table 1. The Four Brands of Cigarette Samples Reported in This Study brand size/length (mm) ISO tar (mg/cig) “intense” tar (mg/cig) A B C D

king (85) king (85) regular (72) regular (72)

4 9 10 15

25 29 29 33

from each year of the study were analyzed for TSNA using LCMS/MS. The results obtained using four internal standards were compared to those calculated using two internal standards from the same samples. To compare the results to those obtained previously by GC-TEA method (Health Canada T-111), the same sample generation conditions (including smoking machine, smoking regimens, and procedures) were also used for the LC-MS/ MS method. The determination of TSNA in cigarette tobaccos was conducted using three replicates from each brand of each year of the study. The extraction solutions used in the GC-TEA (Health Canada T-309) and LC-MS/MS procedures were different, and therefore, independent extractions were prepared from the same homogenized tobacco samples. Sample Generation and Preparation for Mainstream Smoke (MSS) Analyses. MSS was collected on a 92 mm glass fiber filter pad and analyzed originally by the GC-TEA method as per the Health Canada procedure (T-111).5 For LC-MS/MS analysis, the MSS was also collected on the 92 mm pad but only five cigarettes were needed per observation for ISO conditions and three cigarettes for the intense conditions due to the higher sensitivity of the new method. After the MSS was collected, the pad was spiked with 400 µL of the internal standard solution (NNN-d4, NAT-d4, NAB-d4, and NNK-d4) and transferred to a 125 mL amber flask, in which 40 mL of 100 mM ammonium acetate solution was added subsequently. The flask was then agitated for 30 min on a wrist action shaker. Finally, the sample was filtered into an autosampler vial through a 0.45 µm PTFE syringe filter and analyzed by LC-MS/MS. Sample Preparation for Cigarette Tobacco Analyses. GCTEA analysis for cigarette tobacco was performed as per the Health Canada procedure (T-309)5 and required 1 g of homogenized sample for each replicate sample. The LC-MS/MS method required only 0.75 g of sample. The sample was spiked with 300 µL of the internal standard solution and then extracted with 30 mL of 100 mM ammonium acetate solution and agitated for 30 min on a wrist action shaker. Finally, it was filtered into an autosampler vial using a syringe filter and analyzed by LC-MS/ MS. Sample Analysis. An Agilent 1100 series HPLC system coupled with an AB/MDS Sciex API 3000 mass spectrometer was used for sample analysis. Waters XTerra MS C18 column (2.1 mm × 50 mm, 2.5 µm particle size) was used for analyte separation. HPLC conditions were column oven, 70 °C; injection volume, 5 µL; sample-wash, methanol. Mobile phase: A, 0.1% acetic acid in water; B, 0.1% acetic acid in methanol. HPLC separation was achieved using a gradient with mobile phase B increased from 5% at 0 min to 90% at 1 min and then reached 100% at 4 min. The total run time for each sample including equilibration time was

13.5 min. Mass detection conditions were as follows: ionization mode, positive ESI; ion spray voltage, 1500 V; ion source temperature, 450 °C; nebulizer gas, nitrogen, setting, 10; curtain gas, nitrogen, setting, 10; collision gas (CID), nitrogen, setting, 10. Compound-dependent parameters were optimized using flow injection analysis, (FIA) and the optimized values were used: dwell time, 100 ms; DP, 30 V; FP, 200 V; EP, 4 V; CE, 15 eV; CXP, 13 V; and IQ2, -18 V. For each analyte, two ion transition pairs were used under multiple reaction mode (MRM). These ion pairs are 178/148 and 178/120 for NNN, 190/160 and 190/106 for NAT, 192/162 and 192/133 for NAB, and 208/122 and 208/106 for NNK. For internal standards: 182/152 for NNN-d4, 194/164 for NATd4, 196/166 for NAB-d4, and 212/126 for NNK-d4. RESULTS AND DISCUSSION Determination of TSNA in Cigarette Tobacco. For TSNA analysis in cigarette tobacco, the improved LC-MS/MS is more sensitive than the GC-TEA method (as shown in Table S-1 in Supporting Information (Supporting Information)). This increase in sensitivity eliminates the need for concentration steps and simplifies the extraction process. The results obtained for Canadian cigarette tobaccos by the improved LC-MS/MS method are in good agreement with those obtained by the GC-TEA method (as shown in Table S-2 in the Supporting Information) with the exception of NAB, where the values obtained by the GC-TEA method are near the method limit of detection (LOD). The trace amounts of NAB determined by LC-MS/MS show a trend of decrease in the NAB content over the study period (but this trend could not be observed by the GC-TEA method), which is consistent with the trends observed for other TSNA found at higher concentrations, clearly demonstrating the higher sensitivity and accuracy of the LC-MS/MS method for NAB over the GCTEA method. These data also show that the levels of TSNA in commercial Canadian cigarettes have decreased from 2003 to 2005 to the point where all TSNA are near (or below) the limit of quantification (LOQ) for the GC-TEA method. This illustrates a major reason for the development and validation of this improved sensitive method. Determination of TSNA in Mainstream Cigarette Smoke (MSS). For MSS determinations (as shown in Table S-3 in the Supporting Information), the sensitivity of the LC-MS/MS compared to GC-TEA method was increased by a factor of 5.6 for NAB to 23.4 for NNK. The results determined by the two methods for the tested samples are consistent for all TSNA compounds under both ISO and Canadian intense smoking regimens (as shown in Figure S-1). With the GC-TEA method, NAB amounts fall below the LOD under the “ISO” smoking conditions for the final year of the study and are near (or below) the LOQ for all other conditions. These results demonstrate again the need for the higher sensitive LC-MS/MS method developed in this study. LC-MS/MS Methods: Four Internal Standards vs Two Internal Standards. Sample matrix effect is a major concern in development of a LC-MS/MS method.6-10 In this study, sample matrix effect was initially studied by spiking a certain amount of the four internal standards into a standard solution (a clean sample matrix without tobacco) and a real sample matrix (a tobacco extract and a tobacco smoke extract) and then observe their LCMS/MS response changes in the different matrixes. The results

Table 2. Results of Recovery Studies with the KY2R4F Cigarette LC-MS/MS with four ISa LFM (%)

LC-MS/MS with two ISa

analyte

LFB (%)

LFB (%)

LFM (%)

NNN NAT NAB NNK

100 (1.6)b 98.2 (0.9) 98.4 (2.1) 106 (2.3)

Cigarette Tobacco 109 (7.6) 101 (1.7) 102 (5.7) 100 (1.2) 90.8 (6.2) 97.5 (2.3) 112 (2.3) 108 (2.5)

102 (8.1) 219 (7.9) 120 (6.8) 105 (3.7)

NNN NAT NAB NNK

101 (1.7) 110 (3.4) 97.8 (3.1) 112 (3.0)

Mainstream Smoke 99.5 (7.3) 103 (2.1) 104 (5.3) 98.9 (4.1) 99.4 (6.7) 100 (2.8) 110 (5.9) 110 (2.2)

106 (7.8) 235 (8.3) 177 (7.4) 102 (6.3)

a IS stands for internal standards. b Values in parentheses are standard deviations.

(Table S-4 in Supporting Information) show that the responses of the four internal standards in the tobacco matrixes were decreased significantly compared to those from the standard solution, indicating the existence of “ion suppression” matrix effect on the TSNA analysis. In addition, the matrix effect (measured as a response decrease in tobacco matrixes compared to a standard solution) is different for each analyte with a signal reduction of 78% for NNN-d4, 47% for NAT-d4, 59% for NAB-d4, and 79% for NNKd4 in a KY2R4F reference smoke sample, indicating that the matrix effect is analyte-dependent. Because the relative response of NNKd4 in a smoke sample was decreased 79%, which is much higher than those of NAT-d4 (47%) and NAB-d4 (59%), quantification of NAT and NAB using NNK-d4 as an internal standard will certainly lead to higher yields of NAT and NAB (i.e., the overestimation of these two compounds) in tobacco smoke samples. Similar results were also obtained for cigarette tobacco samples (as shown in Table S-4) with slightly different values for each internal standard, indicating that the matrix effects are both analyte dependent and matrix dependent. These matrix effects were further confirmed by studying the analyte recoveries from a laboratory-fortified matrix (LFM) and a laboratory fortified blank (LFB). LFB samples were prepared by spiking a certain amount of analytes and internal standards into a buffer solution (without tobacco) that was used for sample extraction. LFM samples were prepared by spiking the same amount of analytes and internal standards into a real tobacco sample (Kentucky reference cigarette) solution. Two sets of results were generated from the same samples: one was calculated using four internal standards (NNN-d4, NNK-d4, NATd4, and NAB-d4) and another was obtained using only two internal standards (NNN-d4 and NNK-d4). As shown in Table 2, the recoveries for all the analytes are close to 100% from the LFB samples no matter whether two or four internal standards were used, while the results for NAT and NAB from LFM samples are likely to be in error when only two internal standards are used for quantification. This is further exemplified in Table 3, which shows the ratio of results calculated using two internal standards versus four internal standards. The data from Tables 2 and 3 clearly show that the matrix effects on TSNA analysis are analytedependent, which lead to the overestimation of both NAT and NAB amount when NNK-d4 is used as the internal standard for their quantification. Table 3 also shows that these matrix effects Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1343

Table 3. Ratios of Results of Two Internal Standards Versus Four Internal Standards

cigarette tobacco mainstream ISO mainstream intense

analyte

mean ratio (%)

std dev (%)

min (%)

max (%)

NAT NAB total TSNA NAT NAB total TSNA NAT NAB total TSNA

146 147 118 244 235 155 230 203 151

20.6 20.8 7.5 44.2 71.1 16.8 20.1 29.3 8.5

112 112 102 138 98.3 116 183 86.0 132

191 219 136 334 489 195 279 264 178

are not only analyte-dependent but also matrix-dependent such that the NAT amount in cigarette filler can be over estimated by approximately 45%, while NAT in mainstream smoke can be over estimated by >100%. The same is true for NAB, although the amount of NAB is low compared to other TSNA compounds. The overestimation of NAT alone could result in an overestimation of total TSNA amount by as much as 50% in mainstream smoke. Two Ion Transition Pairs in the Improved Method. The inclusion of the second ion-pair transitions improved substantially the method’s selectivity, accuracy, and robustness.11 A consistent peak area ratio of the second ion transition to the first ion transition between the standards and the samples can be used to examine the purity of the peaks and can also be used for a secondary means of analyte confirmation and quantification. The peak-area ratios for the calibration standards were 0.543 ( 0.009, 0.314 ( 0.010, 0.416 ( 0.042, and 0.227 ( 0.010 for NNN, NAT, NAB, and NNK, respectively. The corresponding peak-area ratios found from mainstream smoke analysis of the KY2R4F cigarettes were 0.532 ( 0.011, 0.341 ( 0.007, 0.413 ( 0.013, and 0.255 ( 0.008. Similar results (see Table S-5) were also obtained for cigarette tobaccos. These results showed that there was no interference for each of the ion transition pair and any ion pair for each analyte could be used for quantification. However, to realize the high sensitivity and accuracy of the method, the first ion-pairs were used for quantification since they provide stronger signal intensity than the second ion-pairs. Comparison of the Results with Those Reported. Table 4 shows a comparison of the results obtained in this study with those reported for MSS. There is a good agreement among the laboratories and methods for NNN and NNK. This is expected given the focus on these two analytes and the use of isotopelabeled internal standards for both of them in the LC-MS/MS methods. The situation is less clear with NAB. The published LCMS/MS methods with two internal standards gave values in the

range of 15-20 ng/cigarette, which are in line with the results reported using the GC-TEA method. However, the GC-TEA method cannot accurately analyze NAB in the samples due to its limited sensitivity. The improved LC-MS/MS method that used NAB-d4 as one of the internal standards gave the lowest NAB value of 11.8 ng/cigarette. This finding implies that other methods may need to be verified for possible interferences or integration errors. The situation is similar with NAT. In this case, our use of the LC-MS/MS method with two internal standards resulted in a high value compared to the results from other laboratories using the same method and the results obtained with the GC-TEA method. Again, the use of the LC-MS/MS method with four internal standards gave the lowest value. Similar results and trends were also obtained for cigarette tobacco analyses (see Table S-6 in the Supporting Information). Experimentally, the only difference between this study and the published ones on using two internal standards for quantifications is the LC-MS systems used. The chromatograms obtained in this study (see Figure S-2) are slightly different (the NNK peak is much sharper than those of NAT and NAB) from the published studies, although the same brand of column was used under the same HPLC conditions. The variations in the results for NAB and NAT between our study and the published work further demonstrate that the method (using two internal standards for quantifications) is not robust (easy to be affected by sample matrix effects, small changes in method parameters, and instrumentation) and, therefore, need to be improved. The most recent collaborative study on TSNA analysis in tobacco and tobacco related materials used three internal standards (NNN-d4, NAT-d4, and NNK-d4) to improve the reproducibility of the results between different laboratories.4 CONCLUSIONS One objective of this study is to minimize the sample matrix effect on TSNA analysis without extensive sample cleanup steps. Our initial investigation on sample matrix effect has showed that this matrix effect (mainly ion suppression) is analyte dependent, which is further confirmed by the laboratory fortified sample experiments. To reduce the matrix effects in TSNA analysis, an individual isotope-labeled internal standard was used in this study for calibration of each analyte. The second objective of this study is to achieve analyte confirmation and quantification with greater accuracy and precision. This goal was realized using two ion-pair transitions for each analyte in the sample analysis. On the basis of these improvements, a new LC-MS/MS method has been validated for the determinations of NNN, NAT, NAB, and NNK in cigarette tobacco and mainstream cigarette smoke. This method has demonstrated greater sensitivity, selectivity, and accuracy over the GC-TEA method, especially for NAB, which can only be

Table 4. Comparison of Results for TSNA in ISO MSS of KY2R4F Reference Cigarettes analyte

GC-TEAa ng/cig

GC-TEAb ng/cig

LC-MS/MSc ng/cig

LC-MS/MSd ng/cig

LC-MS/MSe ng/cig

LC-MS/MSf ng/cig

NNN NAT NAB NNK

149 (14) 133 (9.0) 16.1 (1.2) 129 (10)

147 (11) 136 (4.0) 17.0 (1.2) 125 (5.0)

152 (4.0) 112 (5.0) 11.8 (0.8) 133 (6.0)

158 (3.0) 275 (15) 20.7 (1.2) 126 (6.0)

142 (11) 125 (5.0) 17.3 (0.7) 121 (4.0)

155 (17) 122 (8.0) 15.3 (2.0) 134 (10)

a This study using Health Canada method. b Data reported by Lee et al. (ref 3). c This study using four internal standards. d This study using two internal standards as reported by Wagner et al. (ref 2). e Data reported by Lee et al. (ref 3). f Data reported by Wagner et al. (ref 2). g Values in parentheses are standard deviations.

1344 Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

measured accurately by the LC-MS/MS method. With the use of four internal standards to compensate for the matrix effects and reduce the variations during sample preparation and analytical procedures, the improved method has proven to be more accurate and robust compared to the published LC-MS/MS methods. SUPPORTING INFORMATION AVAILABLE LOD and LOQ for determination of TSNA in cigarette tobacco (Table S-1); comparison of the results for TSNA in cigarette tobacco (Table S-2); LOD and LOQ for determination of TSNA in mainstream cigarette smoke (Table S-3); internal standard response decrease caused by sample matrix effects (Table S-4); peak

area ratio of the second ion pair to the first ion pair for cigarette tobacco (Table S-5); comparison of results for TSNA in cigarette tobacco KY2R4F (Table S-6); correlation between the results obtained by GC-TEA and those by LC-MS/MS for MSS sample analysis (Figure S-1); and chromatograms of a mainstream smoke sample obtained in this study (Figure S-2). This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review November 28, 2007.

October

11,

2007.

Accepted

AC702100C

Analytical Chemistry, Vol. 80, No. 4, February 15, 2008

1345