A comparison of the mainstream smoke chemistry and mutagenicity of a representative sample of the US cigarette market with two Kentucky reference cigarettes (K1R4F and K1R5F)

Food and Chemical Toxicology 38 (2000) 949±962

www.elsevier.com/locate/foodchemtox

Research Section A comparison of the mainstream smoke chemistry and mutagenicity of a representative sample of the US cigarette market with two Kentucky reference cigarettes (K1R4F and K1R5F) T. A. CHEPIGA, M. J. MORTON, P. A. MURPHY, J. T. AVALOS, B. R. BOMBICK, D. J. DOOLITTLE, M. F. BORGERDING and J. E. SWAUGER * R.J. Reynolds Tobacco Company, Bowman Gray Technical Center, PO Box 1487, Winston-Salem, NC 27102-1487, USA (Accepted 16 April 2000) AbstractÐThe incorporation of technologies into cigarettes such as ®lters, ®lter ventilation, porous cigarette papers, expanded tobacco and reconstituted tobacco sheet has resulted in cigarettes with a wide range of ``tar'' yields. The objectives of this study were to characterize the US cigarette market according to ``tar'' category (i.e. full ¯avor, FF; full ¯avor low tar, FFLT; or ultra low tar, ULT) and to determine whether the Kentucky reference cigarettes K1R4F and K1R5F are representative of FFLT and ULT cigarettes, respectively. As a means of characterization and comparison, the mainstream smoke from a representative sample of commercially available cigarettes from each market segment and the K1R4F and K1R5F Kentucky reference cigarettes was analyzed for the presence and level of 18 selected chemical constituents. In addition, a measure of the mutagenic activity of the mainstream smoke condensate from these cigarettes was determined using an Ames Salmonella mutagenicity assay. All cigarettes were smoked according to US Federal Trade Commission (FTC) guidelines. Results indicated that, overall, mainstream smoke constituent levels are well predicted by FTC ``tar'' yieldÐconstituent levels increased as ``tar'' delivery increased. Based on the selected analytes measured in mainstream smoke, the K1R4F reference cigarette was generally representative of the FFLT segment of the US cigarette market. The K1R5F reference cigarette was representative of the ULT segment of the US cigarette market for cigarettes with ``tar'' deliveries approximate to it. In terms of mutagenic activity, a direct relationship was also demonstrated on a per cigarette basisÐrevertants per cigarette increased with increasing ``tar'' delivery. There was a weak tendency (R-square=0.12, P=0.08) for speci®c activity (revertants/mg ``tar'') to increase with decreasing ``tar'' yieldÐlower ``tar'' products had a slightly higher speci®c activity. No signi®cant di€erences (P>0.05) were observed when the speci®c activities of the condensates from the K1R4F and K1R5F reference cigarettes were compared to the market segments that they were designed to represent, FFLT and ULT, respectively. Overall, these results support the use of the K1R4F and the K1R5F as acceptable reference cigarettes for comparative mutagenicity and smoke chemistry studies of cigarettes available on the US market. # 2000 Published by Elsevier Science Ltd. Keywords: cigarette smoke; mainstream smoke chemistry; cigarette smoke condensate; cigarette smoke condensate mutagenicity; Kentucky reference cigarettes.

INTRODUCTION

Cigarette design has evolved over the last several decades with the incorporation of many new technologies. Modi®cations to cigarette design have involved the incorporation of new tobacco processes, *Corresponding author.

papers, ®lters and ¯avoring ingredients which either alone or in combination have the potential to modify the quantity and/or the quality of the smoke yielded from the cigarette. The incorporation of these new technologies and materials has resulted in the development a diverse market of commercial cigarettes encompassing a wide range of ``tar'' yields. Collectively, these technologies interact within the complex system of the burning cigarette and, ultimately,

0278-6915/00/$ - see front matter # 2000 Published by Elsevier Science Ltd. PII: S0278-6915(00)00086-7

950

T. A. Chepiga et al.

in¯uence the mainstream smoke formed during the smoking process. During the last 30 to 40 years, a great deal of information has been published regarding the identity of chemicals commonly observed in mainstream cigarette smoke (see Ho€mann, 1993; IARC, 1985; US Surgeon General, 1989). Over this time, signi®cant changes in cigarette design (e.g. air dilution) and tobacco processes, and reductions in ``tar'' have been incorporated by manufacturers. In addition, the activity of mainstream cigarette smoke condensate in in vitro genotoxicity assays has been addressed in the scienti®c literature (DeMarini, 1981, 1983). However, few studies have attempted to concurrently characterize the smoke chemistry and/or the biological activity of the commercial cigarettes presently available on the US market. Recently, members of both the public health and regulatory communities have expressed interest in such data. As cigarette design has continued to evolve, comparative chemical and biological studies of mainstream smoke have formed the foundation of product evaluation strategies adopted to characterize the potential impact of product modi®cation. Owing to the complexity of the cigarette system (attributable to physical cigarette design parameters, variations in tobacco blends and ingredients), the selection of appropriate reference cigarettes for use in comparative product stewardship studies can be challenging. Fortunately, a series of cigarettes has been developed jointly by the US National Cancer Institute, the Agriculture Research Service of the US Department of Agriculture, and the University of Kentucky Tobacco and Health Research Institute (Lexington, Kentucky) to serve as reference cigarettes for experimental purposes (Davis et al., 1984; Sullivan, 1984). These cigarettes, generally referred to as ``Kentucky reference cigarettes,'' incorporate the principal cigarette design advances that have been introduced onto the market since the 1950s. The intention in developing these cigarettes was to provide standard reference cigarettes for use in comparative chemical and biological studies of cigarette smoke. As reference cigarettes, they also provide a basis for comparing data collected in di€erent laboratories and at di€erent points in time. The K1R4F and K1R5F are two Kentucky reference cigarettes that incorporate the design features typical of low ``tar'' cigarettes. As such, standard smoke component yieldsÐfor example ``tar'' (total particulate matter minus nicotine and water), nicotine and carbon monoxideÐfor the K1R4F and K1R5F are typical of low and ultra low tar cigarette designs, respectively. The K1R4F cigarette incorporates ®ltration, air dilution, and a tobacco blend that includes reconstituted tobacco sheet to achieve smoke yields typical of a low tar cigarette. The K1R5F cigarette incorporates expanded or ``pu€ed'' tobacco into the blend, in addition to the design technologies included in the K1R4F cigarette, to

produce smoke yields typical of an ultra low tar cigarette. The K1R4F, developed in 1983, is an 84 mm length, 25 mm circumference cigarette with a cellulose acetate ®lter that yields, on average, 9.2 mg ``tar'' and 0.8 mg nicotine when smoked under standard Federal Trade Commission (FTC) conditions (35 ml pu€ volume, 2 sec duration, 1 pu€/min, smoked to a butt length within 3 mm of the edge of the tipping paper for ®ltered cigarettes and a butt length of 23 mm for un®ltered cigarettes). Based on the tobacco blend, ``tar'', nicotine and carbon monoxide yields (in relation to the US sales weighted average), the K1R4F was designed to be representative (on average) of the US cigarette market, in general, and the full ¯avor low tar (FFLT) segment, in particular. The K1R5F, developed in 1989, is an 84 mm length, 25 mm circumference cigarette with a cellulose acetate ®lter that yields, on average, 1.7 mg ``tar'' and 0.2 mg nicotine when smoked under standard FTC conditions. Based on the tobacco blend, ``tar'', nicotine and carbon monoxide yields, the K1R5F was designed as a reference cigarette for the ultra low tar (ULT) segment of the US cigarette market. In 1995, Steele et al. compared the mutagenicity of mainstream cigarette smoke condensates from a representative sample of the US cigarette market with a Kentucky reference cigarette (the K1R4F). These investigators demonstrated that the K1R4F, as assessed by the Salmonella assay, was a representative model for the US cigarette market using mainstream smoke condensates. To date, the mutagenicity of the K1R5F has not been evaluated in relation to ultra low tar cigarettes currently available on the US market. Further, to date, no reported study has been published which compares the levels of mainstream smoke analytes of commercial cigarettes available on the US market with either the K1R4F or the K1R5F. Findings from this study expand on and complement the earlier work of Steele et al. (1995) in the area of mutagenicity, and also address the current interest of the health and regulatory community concerning more extensive information about mainstream smoke chemistry yields from cigarettes marketed today. The speci®c objectives of this study were twofold: (1) to characterize the US cigarette market in terms of mainstream smoke chemistry and mainstream smoke condensate mutagenicity; and (2) to determine whether the Kentucky reference cigarettes K1R4F and K1R5F are representative of commercial FFLT and ULT cigarettes, respectively.

MATERIALS AND METHODS

Cigarette sample For sample selection, the 1995 US cigarette market was initially strati®ed into 12 di€erent market sec-

A comparison of the mainstream smoke chemistry and mutagenicity

tions: ®rst, according to ``tar'' category (full ¯avor, FF; full ¯avor low tar, FFLT; or ultra low tar, ULT), then according to menthol inclusion (menthol or non-menthol), and ®nally according to price (full price or savings). In stratifying according to ``tar'' category, package self-designation was the primary criterion (typically, the standard ``tar'' delivery range for each of the categories is: ULT, 0±6.5 mg/cigarette; FFLT, >6.5±14.5, and FF, >14.5). Using internal market tracking information, the proportion of market share in each of the 12 strata was then estimated. Because of the extremely low market share of one of the strata, ULT menthol savings, it was combined with the ULT menthol full price stratum prior to the selection process. 25 brands were then randomly selected using a sampling technique described by Cochran, 1977, in which the selection probability increased with increased market share. In order to broaden representation of the ULT ``tar'' category in the resultant sample, four additional ULT brands were added. The ®nal market sample consisted of 29 brands (according to ``tar'' category: 10 FF, 11 FFLT and eight ULT). For the purposes of testing, all market cigarettes were purchased at retail. The K1R4F and K1R5F reference cigarettes were obtained from the University of Kentucky Tobacco and Health Research Institute in Lexington, Kentucky. Preparation of mainstream smoke condensate All cigarettes selected for this study contained ®lters and were smoked according to standard Federal Trade Commission (FTC) conditions (35 ml pu€ volume, 2 sec duration, 1 pu€/min, smoked to a butt length within 3 mm of the edge of the tipping paper) using a 20-port automated smoking machine. K1R4F and K1R5F cigarettes were smoked to a 35 mm butt length. Mainstream particulate matter for each brand was collected on Cambridge ®lter pads. Mainstream smoke constituent analyses The mainstream cigarette smoke from each brand and the two Kentucky reference cigarettes was analyzed for 18 speci®c mainstream smoke constituents. Standard FTC measurements Tar'' and nicotine were determined according to the methodology adopted by the FTC and published by Pillsbury et al. (1969). The ``tar'' content of the cigarette smoke, also referred to as FTC ``tar,'' is de®ned as the weight of the total particulate matter (TPM) collected on the Cambridge pad minus the weight of water and nicotine present in the TPM. Water and nicotine were determined by extracting the pad with 2-propanol, and analyzing the extract using gas chromatography. Carbon monoxide, a vapor or gas phase component of smoke, was determined using non-dispersive infrared (NDIR) spectroscopy after collecting the vapor phase of the mainstream smoke using the method described by Horton and Guerin (1974).

951

Selected mainstream smoke constituent analyses Ammonia was determined colorimetrically according to the method of Harrell et al. (1975). Benzo[a]pyrene was determined by HPLC with ¯uorescence detection (Dumont et al., 1993). Nitrogen oxides were determined by chemiluminescence according to the method of Neurath et al. (1976). Four carbonyl compounds: formaldehyde, acetaldehyde, acetone and acrolein, were determined by liquid chromatography with ¯uorescence detection after derivatization with 2-diphenylacetyl-1,3-indadione-1-hydrazone to form azine derivatives as reported by Borgerding et al. (1997). Hydrogen cyanide was determined by trapping the compound on ascarite followed by water extraction and colorimetric end determination (Collins et al., 1970). Four hydroxybenzenesÐhydroquinone, catechol, phenol, and p-+m-cresolÐwere determined using a modi®cation to the method of Risner and Cash (1990), which employed ®xed detector wavelengths (excitation: 274 nm, emission: 298 nm). Three tobacco-speci®c nitrosamines (TSNAs)ÐN-nitrosonornicotine (NNN), N-nitrosoanatabine (NAT) and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK)Ð were determined using a re®nement of published procedures (Caldwell and Conner, 1990; Ho€man et al., 1984). TSNAs were extracted from mainstream smoke particulate matter with methylene chloride and the extract was passed through a basic alumina column at atmospheric pressure. The resulting solution was concentrated and analyzed by gas chromatography with thermal energy analysis (TEA) detection. Ames Salmonella/microsome plate assays Mainstream cigarette smoke condensate from each of the brands was collected and assayed individually. Condensate was prepared in dimethyl sulfoxide (DMSO) as described previously by Doolittle et al. (1990). Mainstream particulate matter (70±100 mg) from each brand and the two Kentucky reference cigarettes was collected on separate Cambridge ®lter pads. The pads were extracted with DMSO, and the extracted samples were stored at ÿ70 C until assayed. Mutagenicity was assessed in S. typhimurium strains TA98 and TA100 (+S9 metabolic activation) at 0, 25, 50, 75, 100, 125 and 250 mg condensate/plate according to the method of Maron and Ames (1983) with the preincubation modi®cation described by Yahagi et al. (1975). The S9 liver homogenate fraction was prepared according to Ames et al. (1975) from male Sprague±Dawley rats that were administered a single 500 mg/kg ip injection of Aroclor 1254. The S9 concentration in the S9 mix was 5% (v/v) and was added at a volume of 0.5 ml per plate. The S9 mix, the test bacteria, and the test sample were added to a 13 mm85 mm testtube. The mixture was shaken and allowed to incubate for 20 min at 37 C prior to the addition of 2 ml molten top agar supplemented with 0.5 mm histidine/ biotin. The contents of the tube were poured onto

952

T. A. Chepiga et al.

minimal glucose agar and incubated at 37 C for 48 hr. All testing was done using triplicate plates at each concentration. Assays for the US brands and Kentucky reference condensates were conducted concurrently over a period of 6 months. Statistical methods Mainstream smoke constituent analyses The 18 speci®c chemical constituents were measured a total of three to ®ve times in the mainstream smoke from each of the 29 market brands and the two Kentucky reference cigarettes. Arithmetic averages were calculated for each constituent in the mainstream smoke from each of the cigarettes tested. Constituent averages for each ``tar'' category (ULT, FFLT or FF) were then calculated two di€erent ways: (1) As a ®xed ®nite population using a slight modi®cation to the Horvitz±Thompson estimator (Sarndal et al., 1992) where variance was estimated using a slight modi®cation to the Yates±Grundy variance estimator (Cochran, 1977) and comparisons were made using a z-test. This method takes into account both the relative market share of each brand and the probability of each brand's selection in the sample (i.e. provides a sales-weighted approach). (2) As a random sample from a (potentially) in®nite population by taking the simple arithmetic average of all of the brands included in a particular ``tar'' category (i.e. unweighted by market share). Comparisons of mainstream constituent smoke levels from the Kentucky reference cigarettes to the market cigarette levels were also made using a linear regression model relative to the ``tar'' delivery. Ames Salmonella/microsome plate assays Estimates of mutagenicity on a revertant/mg ``tar'' basis were obtained for all assays using the linear regression method of Bernstein for point rejection (Bernstein et al., 1982). Revertants/cigarette were calculated by multiplying the revertants/mg ``tar'' by the mg ``tar''/cigarette. The average mutagenicity of the mainstream smoke condensate from the ULT and FFLT market categories was compared to that of the K1R5F and K1R4F, respectively, as described in the paragraph above, namely: (1) as a ®xed ®nite population; and (2) as a random sample from a (potentially) in®nite population. Comparison of mainstream smoke mutagenic activity (both on a revertant/mg ``tar'' and a revertant/mg cigarette basis) of the Kentucky reference cigarettes to the market cigarette levels was also made using a linear regression model relative to the ``tar'' delivery.

RESULTS

Mainstream smoke constituent analyses Tables 1, 2 and 3 contain the results of the mainstream smoke constituent analyses according to ``tar''

category for the 18 analytes examined. Table 1 contains the results for the market cigarettes in the full ¯avor low tar (FFLT) category and also includes the results from the Kentucky reference cigarette, K1R4F. Table 2 contains the results for the market cigarettes in the ultra low tar (ULT) category and also includes the results from the Kentucky reference cigarette, K1R5F. Table 3 contains the results for the market cigarettes in the full ¯avor (FF) category. In each table, market cigarette results are reported as the unweighted arithmetic average of all cigarettes per ``tar'' category  standard error (column 1), the unweighted range (column 2), the estimated salesweighted average of all cigarettes per ``tar'' category  standard error (column 3), and the estimated salesweighted range (column 4). In general, estimated sales-weighted averages for market cigarettes in each ``tar'' category were similar or slightly higher than unweighted averages for the constituents examined. Without exception, market cigarette averages for each of the 18 constituents in each of the three ``tar'' categories increased with increasing ``tar'' delivery (FF>FFLT>ULT). In Table 1, FFLT market cigarettes were compared to the K1R4F. Statistical analysis (P>0.05) indicated that, for the majority of analytes examined on a sales-weighted basis, levels in the smoke from FFLT cigarettes were similar to levels in the smoke from the K1R4F. In Table 2, ULT market cigarettes were compared to K1R5F. Statistical analysis (P<0.05) indicated that, for the majority of analytes examined on a sales-weighted basis, levels in the smoke from ULT cigarettes were greater than levels in the smoke from the K1R5F. This can most likely be attributed to the di€erence in the average ``tar'' delivery of the ULT market cigarettes compared to the K1R5F (a threefold average di€erence). In fact, one of the market brands included in the ULT category, because it was labeled ``Ultra Light'', actually had a ``tar'' delivery between 7 and 8 mg. According to the ``tar'' category delineation described above, this product would be functionally classi®ed as a FFLT brand. Furthermore, three other of the eight brands classi®ed as ULT in the study delivered >4.5 mg ``tar'', while the K1R5F delivered only 1.5 mg ``tar''. Mainstream smoke constituent analysis results for each individual market cigarette tested are contained in Appendix A. Figure 1 contains plots of the level of each analyte (other than ``tar'') measured in mainstream smoke according to ``tar'' delivery for all cigarettes examined. From these graphs, it is apparent that there is a linear relationship between mainstream smoke analyte delivery and ``tar'' delivery (R-square values ranging from 0.70 to 0.96). That is, across all ``tar'' categories, mainstream smoke analyte delivery increases with increasing ``tar''. For the K1R4F, these graphs illustrate its usefulness as being representative of the US FFLT cigarette market. For the K1R5F, these graphs illustrate its usefulness as being representative

A comparison of the mainstream smoke chemistry and mutagenicity

953

Table 1. Mainstream smoke constituent analyses: full ¯avor low tar (FFLT) brands and the Kentucky reference cigarette K1R4F (analyte/ cigarette) Analyte ``Tar'' (mg/cig) Nicotine (mg/cig) Carbon monoxide (mg/cig) Ammonia (mg/cig) Benzo[a]pyrene (ng/cig) Nitrogen oxides (mg/cig) Formaldehyde (mg/cig) Acetaldehyde (mg/cig) Acetone (mg/cig) Acrolein (mg/cig) Hydrogen cyanide (mg/cig) Hydroquinone (mg/cig) Catechol (mg/cig) Phenol (mg/cig) p-/m-Cresol (mg/cig) NNN (ng/cig) NAT (ng/cig) NNK (ng/cig)

Unweighted average

Unweighted range

Estimated sales-weighted average

Estimated sales-weighted range

K1R4F

9.900.40 0.760.03 11.50.8 19.81.6 6.300.30 22519 16.01.5 64945 26019 63.34.3 16816 41.71.6 44.71.8 12.11.4 8.300.60 87.08.0 1008.0 81.05.0

8.10±11.6 0.62±0.94 5.40±14.4 12.9±28.2 5.15±7.50 80.8±290 10.4±24.7 342±812 116±310 43.1±79.3 59.6±237 34.7±48.7 38.5±52.7 6.25±21.5 5.84±11.8 53.0±142 73.0±159 62.0±98.0

10.00.4a 0.770.03 11.70.6 20.91.1 6.300.20a 23017 15.51.0a 66343 26417 65.53.4 17511 41.41.6 44.61.8 11.60.7a 8.000.40 94.06.0a 1067.0 79.04.0

8.10±11.6 0.60±0.94 5.40±14.4 12.9±28.2 5.20±7.5 81.0±290 10.4±24.7 342±812 116±310 43.1±79.3 60.0±237 34.7±48.7 36.6±52.7 6.30±21.5 5.80±11.8 53.0±142 73.0±159 62.0±115

8.70 0.78 11.0 18.8 4.60 266 12.5 707 284 59.7 144 39.8 43.5 9.60 7.20 67.0 93.0 78.0

a

Statistically signi®cantly di€erent than K1R4F (P<0.05).

Table 2. Mainstream smoke constituent analyses: ultra low tar (ULT) brands and the Kentucky reference cigarette K1R5F (analyte/ cigarette) Analyte ``Tar'' (mg/cig) Nicotine (mg/cig) Carbon monoxide (mg/cig) Ammonia (mg/cig) Benzo[a]pyrene (ng/cig) Nitrogen oxides (mg/cig) Formaldehyde (mg/cig) Acetaldehyde (mg/cig) Acetone (mg/cig) Acrolein (mg/cig) Hydrogen cyanide (mg/cig) Hydroquinone (mg/cig) Catechol (mg/cig) Phenol (mg/cig) p-/m-Cresol (mg/cig) NNN (ng/cig) NAT (ng/cig) NNK (ng/cig)

Unweighted average

Unweighted range

Estimated sales-weighted average

Estimated sales-weighted range

K1R5F

3.300.90 0.310.07 4.801.10 7.201.70 2.500.60 11221 3.600.90 23447 99.019.0 17.94.5 44.013.0 15.54.1 17.54.7 4.202.00 2.901.10 45.09.0 52.011.0 40.010.0

0.70±7.20 0.11±0.59 1.70±8.90 2.63±16.2 0.79±5.49 45.9±183 1.20±7.60 96.1±389 41.7±161 4.30±32.5 13.0±110 3.56±35.1 4.02±40.8 0.66±17.2 0.54±10.2 18.0±96.0 25.0±110 13.0±83.0

4.500.30a 0.400.03a 5.800.50 9.200.60a 3.400.20a 1398 4.600.30a 28222a 1189a 23.42.0a 43.06.0 19.71.3a 22.61.5a 6.000.30a 4.200.20a 55.03.0a 63.04.0a 52.04.0a

0.70±7.20 0.11±0.59 1.70±8.90 2.60±16.2 0.79±5.50 45.9±190 1.20±7.60 96.1±397 41.7±160.7 4.30±32.5 13.0±109.7 3.60±35.1 4.00±40.8 0.66±17.2 0.54±10.2 18.0±96.0 25.0±110 13.0±83.0

1.50 0.18 3.00 5.30 1.50 96.0 1.80 156 70.0 9.50 25.0 5.80 8.10 0.90 0.70 29.0 39.0 18.0

a

Statistically signi®cantly di€erent than K1R5F (P<0.05).

Table 3. Mainstream smoke constituent analyses: full ¯avor (FF) brands (analyte/cigarette) Analyte ``Tar'' (mg/cig) Nicotine (mg/cig) Carbon monoxide (mg/cig) Ammonia (mg/cig) Benzo[a]pyrene (ng/cig) Nitrogen oxides (mg/cig) Formaldehyde (mg/cig) Acetaldehyde (mg/cig) Acetone (mg/cig) Acrolein (mg/cig) Hydrogen cyanide (mg/cig) Hydroquinone (mg/cig) Catechol (mg/cig) Phenol (mg/cig) p-/m-Cresol (mg/cig) NNN (ng/cig) NAT (ng/cig) NNK (ng/cig)

Unweighted average

Unweighted range

Estimated sales-weighted average

Estimated sales-weighted range

15.40.4 1.100.05 16.10.7 33.42.3 9.100.20 30012 27.72.0 91628 35111 1005 28313 59.91.5 62.82.2 18.11.1 12.30.6 15010 17214 13511

12.9±17.2 0.79±1.39 12.1±19.9 23.1±47.6 7.96±9.97 244±367 21.0±40.3 726±1071 287±393 75.4±122 229±359 52.2±67.2 49.4±73.4 11.12±23.7 9.29±15.9 93.0±191 108±235 98.0±212

15.70.3 1.140.03 15.90.6 37.01.3 9.200.20 30314 25.70.7 92022 34712 99.33.0 28710 60.61.8 64.32.0 19.20.8 12.90.5 1587 1829 1318

12.9±17.2 0.79±1.39 12.1±19.9 23.1±47.6 8.00±10.0 244±367 21.0±40.3 726±1071 287±393 75.4±122 229±359 52.2±67.2 49.4±73.4 11.1±23.7 9.30±15.9 93.0±191 108±235 99.0±212

Fig. 1. Plots of the level of each analyte except ``tar'' (y-axis) vs ``tar'' delivery (x-axis) measured in the mainstream smoke of each of the 29 US cigarette brands (open squares) selected as the market sample, and the two Kentucky reference cigarettes (solid squares). Vertical lines delineate ``tar'' categories into ULT, FFLT and FF, respectively.

954 T. A. Chepiga et al.

955

Fig. 1. (continued)

A comparison of the mainstream smoke chemistry and mutagenicity

T. A. Chepiga et al.

Fig. 1. (continued)

956

957

Fig. 1. (continued)

A comparison of the mainstream smoke chemistry and mutagenicity

958

T. A. Chepiga et al.

Fig. 1. (continued)

of the US ULT cigarette market for cigarettes with an approximate ``tar'' delivery (i.e. close to 1.5 mg). Ames Salmonella/microsome plate assays Table 4 contains the results of mainstream smoke condensate mutagenicity testing using the Ames Salmonella/microsome plate assay incorporating strains TA98 and TA100 in the presence of metabolic activation. For each ``tar'' category and the two Kentucky reference cigarettes, results are reported on both a revertant/mg ``tar'' and revertant/cigarette basis. Similar to the mainstream smoke constituent analyses (Tables 1±3), results from each market cigarette ``tar'' category are presented as unweighted, arithmetic averages ( standard error), unweighted ranges, estimated sales-weighted averages ( standard error) and estimated sales-weighted ranges.

In general, results using estimated sales-weighted averages for market cigarettes in each ``tar'' category were similar or slightly higher than unweighted averages. A comparison of the FFLT market ``tar'' category with the K1R4F indicated that there were no statistical di€erences (P>0.05) on either a revertant/mg ``tar'' or a revertant/cigarette basis in either strain of bacteria. Likewise, a comparison of the ULT market ``tar'' category with K1R5F indicated that there were no statistical di€erences (P>0.05) on a revertant/mg ``tar'' basis in either strain of bacteria. On a revertant/ cigarette basis, however, the ULT market ``tar'' category displayed statistically signi®cant greater activity (P<0.05), in both strains of bacteria, compared to the K1R5F. Ames Salmonella/microsome plate assay results for each individual market cigarette tested are presented in Appendix B.

Table 4. Ames Salmonella/microsome plate assay results

Full ¯avor low tar (FFLT)

Ultra low tar (ULT)

Full ¯avor (FF)

a

Bacterial strain

Unweighted average

Unweighted range

Estimated sales-weighted average

Estimated sales-weighted range

TA98 (rev/mg ``tar'') TA98 (rev/cig)a TA100 (rev/mg ``tar'') TA100 (rev/cig)a

132774 133171170 77062 7669737

970±1576 8352±16938 462±1024 3978±10088

139989 14219894 77472 77951414

970±1576 8352±16938 462±1024 3978±10088

K1R4F 1417 13083 852 7716

TA98 (rev/mg ``tar'') TA98 (rev/cig)a TA100 (rev/mg ``tar'') TA100 (rev/cig)a

148078 50451228 85599 2835683

1247±1901 1832±10151 506±1416 764±5675

1247±1901 1832±10151 506±1416 764±5675

K1R5F 1349 2695 730 1163

TA98 (rev/mg ``tar'') TA98 (rev/cig)a TA100 (rev/mg ``tar'') TA100 (rev/cig)a

rev/cig= rev/mg ``tar''mg ``tar''/cig. Statistically signi®cantly di€erent than K1R5F (P<0.05).

b

130638 20093799 77941 11936584

1128±1493 16231±24556 553±1024 9401±16660

144338 69131415b 94654 4116683b 134987 21084648 78878 12267555

1128±1493 16231±24556 553±1024 9401±16660

Fig. 2. Plots of Ames Salmonella mutagenicity results (y-axis) vs ``tar'' delivery (x-axis) measured using the mainstream smoke condensate of each of the 29 US cigarette brands (open squares) selected as the market sample, and the two Kentucky reference cigarettes (solid squares). Vertical lines delineate ``tar'' categories into ULT, FFLT and FF, respectively. In the two upper graphs, mutagenicity results are presented as revertants/mg ``tar'' for each of the cigarette condensates tested using bacterial strains TA98 and TA100, respectively. In the two lower graphs, results are presented as revertants/cigarette.

A comparison of the mainstream smoke chemistry and mutagenicity 959

960

T. A. Chepiga et al.

Figure 2 contains plots of the Ames smoke condensate mutagenicity results in the two strains of bacteria on both a revertant/mg ``tar'' and a revertant/cigarette basis according to ``tar'' delivery for all cigarettes examined. On a revertant/cigarette basis, the two lower graphs in Fig. 2 illustrate that, in both strains of bacteria, mutagenic activity increases as a function of the quantity of ``tar'' delivered. On a revertant/mg ``tar'' basis, the relationship of mutagenic activity to ``tar'' is less clear. The two upper graphs in Fig. 2 illustrate that there is a weak tendency for mutagenic activity, as measured on a revertant/mg ``tar'' basis in both strains of bacteria, to decrease with increasing ``tar'' deliveries. That is, there is a weak tendency for lower ``tar'' products to display a higher Ames activity/mg ``tar'' (R-square=0.12, P=0.08).

DISCUSSION

Many design di€erences can and do exist among di€erent styles of commercially marketed cigarettes. These include the quantity and types of tobacco utilized (including reconstituted tobacco sheet and expanded tobacco); the crop year(s) of the tobacco utilized; the quantity and types of ingredients utilized; the nontobacco materials utilized (i.e. papers and ®lters); and the physical design of the cigarette. These differences in cigarette design have resulted in the development of a diverse market of commercial cigarettes with a wide range of ``tar'' yields. The Kentucky reference cigarette series was designed to provide reference standards representative of US market cigarettes, according to ``tar'' category, for the purposes of comparison testing. Steele et al. (1995) demonstrated that the K1R4F is representative of the US market as measured by mutagenic activity in the Ames Salmonella/microsome plate assay. The results of the present testing further corroborate this ®nding. In terms of the mutagenic activity of mainstream smoke condensate, the K1R4F is representative of commercially marketed US FFLT cigarette brands. Comparative analysis of the K1R5F (Fig. 2) indicates that it is representative of commercially marketed US ULT cigarette brands that have a similar ``tar'' delivery (i.e. close to 1.5 mg). Mainstream cigarette smoke is a complex mixture estimated to contain as many as 4000 individual chemical constituents (Adams et al., 1987; Dube and Green, 1982). This complexity re¯ects the dynamics of the processes of combustion and pyrolysis inherent to the burning tobacco rod. In the present study, 18 mainstream smoke constituents were selected for analysis. The selected constituents have been previously identi®ed by the public health community as compounds which may contribute to the risks associated with cigarette smoking (Ho€mann, 1993; IARC, 1985; US Surgeon General, 1989). Analysis of

the current data clearly demonstrates that the yields associated with the K1R4F reference cigarette are representative of the average yields associated with the FFLT ``tar'' segment of the US cigarette market. Although reasonably similar, the relationship between the K1R5F and the ULT market segment is weaker. The apparent weakness of this overall correlation is most likely a re¯ection of the di€erences in ``tar'' delivery between the K1R5F and a number of market cigarettes selected for the ULT segment in this study. Four of the eight ULT brands in the study delivered greater than 4.5 mg ``tar'', while the K1R5F ``tar'' delivery was 1.5 mg. From Fig. 1 it is evident that the other four lower ``tar'' brands in the ULT category, which had similar ``tar'' deliveries to the K1R5F, also displayed similar analyte deliveries compared to the K1R5F. It is reasonable to conclude that the mainstream smoke chemistry of the K1R5F is representative of ULT brands with similar ``tar'' yields. Overall, these results support the use of the K1R4F and the K1R5F as acceptable reference cigarettes for comparative mutagenicity and smoke chemistry studies of cigarettes available in the US marketplace. In this study, there was a closer correlation observed between the K1R4F and the selected FFLT brands than that observed between the K1R5F and the selected ULT brands. This was mostly likely due to the fact that half of the selected brands self-designated as ltra low tar yielded more than three times the ``tar'' delivery of the K1R5F. Perhaps the most compelling results of the present study are the correlations observed between mainstream smoke ``tar'' delivery and mainstream smoke analyte constituent levels. As illustrated in Fig. 1, there is a reasonably strong positive correlation between mainstream smoke analyte constituent level and ``tar'' delivery for every constituent measured± constituent levels increased as ``tar'' delivery increased. It is noteworthy that these relationships exist despite the signi®cant di€erences in design characteristics that are observed in the US marketplace. In short, these results indicate that, overall, mainstream smoke constituent levels are well predicted by the FTC ``tar'' yield. A similar relationship was observed between mainstream smoke ``tar'' delivery and mainstream smoke condensate mutagenic activity. As illustrated in Fig. 2, there is a reasonably strong positive correlation between mainstream smoke condensate activity level on a revertant/cigarette basis and ``tar'' delivery± activity increased as ``tar'' delivery increased. Interestingly, on a revertant/mg ``tar'' basis, there is a weak tendency for the converse. This relationship was also noted by Steele et al. (1995), who reported a negative correlation between speci®c activity (mutagenicity on a revertant/mg condensate basis) and mg condensate/cigarette in the Salmonella assay for 24 US market brands. Steele et al. (1995) also observed a positive correlation between mutagenicity and air

A comparison of the mainstream smoke chemistry and mutagenicity

dilution. They hypothesized that increased air dilution, which generally results in cigarettes with a lower yield of condensate/cigarette, might alter the way the cigarettes burn and could yield a slightly more mutagenic condensate. However, similar to the results observed in the present study, mutagenicity on a per cigarette basis was lower for low condensate products in the Steele et al. (1995) report since total condensate yielded was lower. Also, in an earlier report, Sato et al. (1977) observed that speci®c mutagenic activity (i.e. revertants/mg smoke condensate), as measured by the Salmonella assay, was comparable among smoke condensates from low to high ``tar'' cigarettes (which included eight US brands). Therefore, these authors suggested that reducing the amount of ``tar'' in cigarettes e€ectively decreases the amount of mutagen inhaled from a single cigarette. Collectively, these observations are noteworthy in light of the tremendous level of diversity present in the cigarettes commercially available in the US marketplace. Overall, the results of the present study strongly indicate that both mainstream smoke constituent levels and mutagenic activity on a per cigarette basis are directly related to the amount of ``tar'' delivered per cigarette. In short, despite the di€erences which exist on a by-brand basis, mainstream smoke constituent levels and mutagenic activity on a per-cigarette basis are well predicted by the FTC ``tar'' yield. REFERENCES

Adams J., O'Mara-Adams K. and Ho€mann D. (1987) Toxic and carcinogenic agents in undiluted mainstream smoke and sidestream smoke of di€erent types of cigarettes. Carcinogenesis 8, 729±731. Ames B. N., McCann J. and Yamasaki E. (1975) Methods for detecting carcinogens and mutagens with the Salmonella/mammalian microsome mutagenicity test. Mutation Research 31, 347±364. Bernstein L., Kaldor J., McCann J. and Pike M. C. (1982) An empirical approach to the statistical analysis of mutagenesis data from the Salmonella test. Mutation Research 97, 267±281. Borgerding M., Bodnar J., Chung H., Mangan P., Morrison C., Risner C., Rogers J., Simmons D., Uhrig M., Wendelboe F., Wingate D. and Winkler L. (1997) Chemical and biological studies of a new cigarette that primarily heats tobacco. Part 1. Chemical composition of mainstream smoke. Food and Chemical Toxicology 36, 169±182. Caldwell W. S. and Conner J. M. (1990) Artifact formation during smoke trapping: An improved method for determination of N-nitrosamines in cigarette smoke. Journal of the Association of Ocial Analytical Chemists 73, 783± 789. Cochran W. G. (1977) Sampling Techniques. Wiley series in probability and mathematical statistics. John Wiley & Sons, New York. Collins P. F., Sarji N. M. and Williams J. F. (1970) An automated method for determination of hydrogen cyanide in cigarette smoke. Tobacco Science 14, 12±15. Davis D. L., Vaught A., Tso T. C. and Bush L. P. (1984) Analysis of a new low yield research cigarette. Tobacco and Health Institute, Lexington, KY.

961

DeMarini D. M. (1981) Mutagenicity of fractions of cigarette smoke condensate in Neurospora crassa and Salmonella typhimurium. Mutation Research 88, 363±374. DeMarini D. M. (1983) Genotoxicity of tobacco smoke and tobacco smoke condensate. Mutation Research 114, 59± 89. Doolittle D. J., Lee C. K., Ivett J. L., Mirsalis J. C., Riccio E., Rudd C. J., Burger G. T. and Hayes A. W. (1990) Comparative studies on the genetoxic activity of mainstream smoke condensate from cigarettes which burn or only heat tobacco. Environmental and Molecular Mutagenesis 15, 93±105. Dube M. and Green C. (1982) Methods of collection of smoke for analytical purposes. Recent Advances in Tobacco Science 8, 42±102. Dumont J., Larocque-Lazure F. and Iorio C. (1993) An alternative isolation procedure for the subsequent determination of benzo(a)pyrene in total particulate matter of cigarette smoke. Journal of Chromatographic Science 31, 371±374. Harrell T. G., Rush K. L. and Sensabaugh A. J. Jr (1975) Colorimetric method for the determination of ammonia in tobacco smoke. Tobacco Science 19, 154±156. Ho€mann D. (1993) Analysis of toxic smoke constituents. Report to the US Consumer Product Safety Commission, Chapter D. Ho€mann D., Brunnemann K. D., Adams J. D. and Hecht S. S. (1984) Formation and analysis of N-nitrosamines in tobacco products and their endogenous formation in consumers. In N-Nitroso Compounds: Occurrence, Biological E€ects and Relevance to Human Cancer, eds. I.K. O'Neill, R.C. von Borstel, C.T. Miller, J. Long and H. Bartsch. IARC Scienti®c Publications no. 57, pp. 743± 762. International Agency for Research on Cancer, Lyon. Horton A. D. and Guerin M. R. (1974) Gas-solid chromatographic determination of carbon monoxide and carbon dioxide in cigarette smoke. Journal of the Association of Ocial Analytical Chemists 57, 1±7. IARC (1985) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. Vol. 38. Tobacco Smoking. pp. 83±126. International Agency for Research on Cancer, Lyon. Maron D. M. and Ames B. N. (1983) Revised methods for the Salmonella mutagenicity test. Mutation Research 113, 173±215. Neurath G. B., Dunger M. and Pein F. G. (1976) Interaction of nitrogen oxides, oxygen, and amines in gaseous mixtures. In Environmental N-nitroso Compounds Analysis and Formation. IARC Scienti®c Publications no. 14, ed. E. A. Bogovski, P. Bogovski and L. Griciute, pp. 215± 225. International Agency for Research on Cancer, Lyon. Pillsbury H. C., Bright C. C., O'Connor K. J. and Irish F. W. (1969) Tar and nicotine in cigarette smoke. Journal of the Association of Ocial Analytical Chemists 52, 458± 462. Risner C. H. and Cash S. L. (1990) A high-performance liquid chromatographic determination of the major phenolic compounds in tobacco smoke. Journal of Chromatographic Science 28, 239±244. Sarndal C., Swenson B. and Wretman J. (1992) Model Assisted Survey Sampling. Springer Series in Statistics. Springer-Verlag, New York. Sato S., Seino Y., Ohka T., Yahagi T., Nagoa M., Matsushima T. and Sugimura T. (1977) Mutagenicity of smoke condensates from cigarettes, cigars, and pipe tobacco. Cancer Letters 3, 1±8. Steele R. H., Payne V. M., Fulp C. W., Rees D. C., Lee C. K. and Doolittle D. J. (1995) A comparison of the mutagenicity of mainstream cigarette smoke condensates from a representative sample of the US cigarette market with a Kentucky reference cigarette (K1R4F). Mutation Research 342, 179±190.

962

T. A. Chepiga et al.

Sullivan S. (1984) The Reference and Research Cigarette Series, University of Kentucky Printing Services. Lexington, KY. US Surgeon General (1989) Reducing the Health Consequences of Smoking. DHHS Publ. (CDC) 89-8411.

Yahagi T., Degawa M., Seino Y., Matsushima J., Nagao M., Sugimura T. and Hashimoto Y. (1975) Mutagenicity of carcinogenic azo dyes and their derivatives. Cancer Letters 1, 91±97.