Physical design analysis and mainstream smoke constituent yields of the new potential reduced exposure product, Marlboro UltraSmooth

Transcription

1 Nicotine & Tobacco Research Volume 9, Number 11 (November 2007) Physical design analysis and mainstream smoke constituent yields of the new potential reduced exposure product, Marlboro UltraSmooth Vaughan W. Rees, Geoffrey Ferris Wayne, Brian F. Thomas, Gregory N. Connolly Received 29 June 2006; accepted 21 December 2006 Potential reduced exposure products (PREPs) purport to lower toxicant emissions, but without clinical and longterm health outcome data, claims for reduced harm status of PREPs depend heavily on standard machine yield smoke constituent data. Two prototypes of the new carbon-filtered PREP Marlboro UltraSmooth (MUS) were investigated using both standard (FTC/ISO) and intensive (Health Canada) machine methods to measure gas/ vapor- and particulate-phase smoke constituents. Basic physical design characteristics that may influence smoke constituent yields, such as ventilation, pressure drop (resistance to draw), quantity of tobacco, and quantity and type of carbon, were measured. The possible presence of added chemical flavorant compounds was investigated using gas chromatography mass spectroscopy. MUS prototypes were found to have several key differences in physical design compared with a conventional cigarette, including higher ventilation, lower draw resistance, and in the case of the Salt Lake City prototype, the use of vitreous carbon beads and the presence of chemical flavorants on both the beads and an embedded filter fiber. When tested under the standard regimen, gas-phase constituents of MUS prototypes were reduced compared with a conventional low-yield cigarette. However, far smaller reductions in gas-phase constituents were observed under the intensive regimen, suggesting that the carbon technology used in MUS is less effective when smoked under more intensive conditions. Particulate-phase constituents were not reduced by the carbon filter under either machine-smoking regimen. The data suggest that MUS has been designed to reduce toxic yields while preserving consumer appeal. However, MUS is less effective in reducing toxic smoke constituents when smoked under intensive conditions. Introduction Tobacco manufacturers have begun to introduce products that may produce lowered toxicant emissions when measured under a standard smoke machine test. These so-called potential reduced exposure products (PREPs) could reduce mortality and morbidity associated with tobacco use. However, comprehensive assessment of PREPs has not been Vaughan W. Rees, Ph.D., Geoffrey Ferris Wayne, M.A., Gregory N. Connolly, D.M.D., M.P.H., Division of Public Health Practice, Harvard School of Public Health, Boston, MA; Brian F. Thomas, Ph.D., Center for Chemistry Services, Health Sciences Unit, RTI International, Research Triangle Park, NC. Correspondence: Vaughan Rees, Ph.D., Harvard School of Public Health, Division of Public Health Practice, Landmark Building, Level 3 East, 677 Huntington Avenue, Boston, MA 02115, USA. Tel: +1 (617) ; Fax: +1 (617) ; vrees@hsph.harvard. edu undertaken and the empirical evidence is insufficient to draw conclusions about the potential for PREPs to reduce the harm associated with tobacco consumption. The Institute of Medicine reviewed assessment strategies and indicated areas of research required to evaluate the reduced harm status of PREPs (Stratton, Shetty, Wallace, & Bondurant, 2001). Although an understanding of the dose response relationship between constituent exposure and health outcomes is a long-term research priority, shorter term goals also must be pursued. A key preliminary step is a comparison of toxic mainstream smoke constituent yields of PREPs to those of conventional products. A minimum criterion for a reduced harm product is evidence of lower toxic smoke constituent yields, and the Institute of Medicine has proposed that tobacco products be assessed for yields of ISSN print/issn X online # 2007 Society for Research on Nicotine and Tobacco DOI: /

2 1198 NEW POTENTIAL REDUCED EXPOSURE PRODUCT nicotine and other tobacco toxicants according to a method that reflects actual circumstances of human consumption (Stratton et al., 2001, p. 10). Marlboro UltraSmooth (MUS) is a new PREP manufactured by Philip Morris that uses a modified filter containing activated carbon in the form of charcoal granules. MUS entered a commercial test market in the U.S. cities of Atlanta, Salt Lake City, and Tampa in April In June 2005, another test market was added, in the state of North Dakota, where MUS was sold under the Marlboro Ultra Lights brand. Relatively little information has been made public about the design modifications of MUS compared with a conventional cigarette, and how such changes influence delivery of toxic smoke constituents. Internal tobacco industry documents suggest that MUS is the result of extensive research on carbon filters conducted through the 1990s by Philip Morris (e.g. Philip Morris, 2002). The primary goal of this research was to develop a reduced-exposure tobacco product in the style of a conventional cigarette (Philip Morris, 2001a). Activated carbon may reduce gas/vapor-phase constituents, including the carcinogenic volatile organic and carbonyl groups (Stratton et al., 2001). Carbon has little effect on reducing carbon monoxide, nicotine, and virtually all particulate-phase constituents, including polycyclic aromatic hydrocarbons (PAHs) and tobacco-specific nitrosamines (Hoffman & Hoffman, 1997; Philip Morris, 2002). Internal industry testing of early carbon filter prototypes has demonstrated the same finding (Philip Morris, 2001b, 2002). Consumer studies conducted internally by Philip Morris suggested that carbon-filtered cigarettes reduce or alter flavor, possibly owing to reductions in flavor-carrying volatile compounds (Philip Morris, 2002). These changes could result in reduced consumer acceptance of carbon-filtered cigarettes. Internal documents suggest that Philip Morris sought to improve the consumer acceptability of MUS by adding an artificial flavorant (Philip Morris, 2000, 2001c). A U.S. patent filed in 2004 by Philip Morris (Jupe et al., 2004) described a carbon filter with an added artificial flavor system, which may be used in MUS. According to the patent, the filter delivers flavorant compounds via a chemically impregnated cotton fiber thread inserted into the cellulose acetate filter tow material. The flavorant, which is not revealed in the patent, also may be added to the carbon granules themselves. The present study explored the design characteristics and mainstream smoke constituent reduction potential of MUS. Physical characteristics that may influence smoke constituent yields, including ventilation, pressure drop (resistance to draw), tobacco quantity, and characterization of carbon and carbon quantity, were investigated. The possible inclusion by the manufacturer of added chemical flavorant compounds was explored using chemical analytic techniques. Mainstream smoke yields were assessed using machine-smoking protocols. In addition to tar, nicotine, and carbon monoxide, constituents tested included a range of volatile organic, phenolic, and carbonyl compounds, which include known carcinogens (International Agency for Research on Cancer, 2004). PAHs were included to determine the relative effect of the MUS carbon filters on particulate-phase constituents. Finally, hydrogen cyanide was included because it is an important cardiovascular (noncancer) toxin (Fowles & Dybing, 2003). Although the FTC protocol is the standard assessment strategy required for industry reporting of product mainstream smoke constituents to the United States (Federal Trade Commission, 1980), this method has been criticized for its poor concordance with actual human puffing behavior (e.g., Hammond et al., 2006; Kozlowski, O Connor, & Sweeney, 2000) and for potentially underestimating cigarette smoke yields. This may be particularly true for low tar and nicotine yield conventional cigarettes (Benowitz, Jacob, Kozlowski, & Yu, 1986; Djordjevic, Hoffmann, & Hoffmann, 1997). The FTC method calls for 35-ml puffs of 2 s duration, taken once per minute. The more intensive Health Canada (HC) method calls for 55-ml puffs lasting 2 s, taken every 30 s (Health Canada, 1999). In addition, the HC method requires 100% vent hole blocking, whereas the FTC protocol requires 0% blocking. Mainstream smoke yields for MUS were assessed using both the standard FTC machine-smoking regimen and the intensive HC regimen, to allow comparison of product yields under different smoking conditions. Method Cigarettes Cigarettes investigated were MUS test market brands: Atlanta, Salt Lake City (SLC), Tampa, and North Dakota (ND), and the conventional low-yield Philip Morris product Marlboro Ultra Lights (MUL). The cigarettes were purchased from retail outlets in the respective test market cities and shipped to the study authors in Massachusetts. Physical product analysis General physical features of the whole cigarette were inspected visually, and linear dimensions and weight were measured. The filter of each brand was cut lengthwise with a scalpel to reveal the charcoal

3 NICOTINE & TOBACCO RESEARCH 1199 granules, which were removed from the cellulose acetate filter material. Pressure drop (mmh 2 O) and ventilation (%) were determined using a KC-3 Pressure Drop Ventilation Instrument (Borgwaldt- KC, Richmond, Virginia). Because of the unusual characteristics of the carbon particles, charcoal from MUS-SLC, MUS- Tampa, and MUS-Atlanta was removed from the filter cavity and examined using a scanning electron micrograph (SEM). Samples of the beads/particles were analyzed to observe the exterior surface of the carbon and then crushed to observe the interior structure. Artificial flavor system analysis The flavor system of the MUS-SLC prototype was chosen for further investigation, owing to its unique cotton flavor fiber and the high quantity of bead charcoal, which best match the flavor system described in U.S. Patent No. 6,761,174 (Jupe et al., 2004). The flavor fiber and carbon granules were analyzed using gas chromatography mass spectroscopy (GC-MS) to identify chemical additives. Two Pasteur pipettes were fashioned into columns using glass wool packing. The carbon granules and flavor fiber were placed separately into the two pipette columns. Each column was then washed with 1 ml of acetone, methylene chloride, and methanol, respectively. Eluents were collected individually in concentration vials. The columns were allowed to gravity drip before being dried between solvents using dry nitrogen. The samples were then concentrated under dry nitrogen to a volume of less than 50 ml (i.e., a 20- fold concentration). Sample extracts were analyzed by GC-MS. The system was composed of an HP 5890 GC and a HP 5989A MS. The injection size was 1 ml injected manually. The column used was a J&W Scientific DB-5MS: 40 m, 0.18 mm ID, 0.18 mm film. All samples were run in EI mode with a splitless injection. The inlet temperature was 250uC, and the detector temperature was 280uC. The GC oven temperature programs were as follows: for acetone and methylene chloride samples, the initial oven temperatures were 35uC and 30uC, respectively; the temperature was increased at 60uC/min to 50uC and held for 5 min; the temperature was increased at a rate of 10uC/min to 300uC and held for 5 min. For methanol samples, the starting temperature was 50uC. The temperature was raised to 75uC at a rate of 60uC/min and held for 5 min, at which point the temperature was raised to 300uC at a rate of 10uC/ min and then held for 5 min. The mass spectrometer was tuned using PFTBA and run in Scan mode with an EM voltage of 402 above the autotune value. A solvent delay of 6 min was used. The MS scan ranged from 35 to 400 amu. The source temperature was 200uC, and the quadrupole temperature was set at 100uC. Smoke constituent analyses The two MUS brands with the highest carbon load, SLC and Tampa, were selected for analysis of mainstream smoke constituent yields, and compared with the ultra-light conventional cigarette, MUL. Only two MUS prototypes were analyzed, owing to budgetary constraints. All brands were tested using both the standard FTC and the intensive HC machine-smoking protocols (FTC, 1980; Health Canada, 1999). In addition to tar, nicotine, and carbon monoxide, the analyte groups assessed were gas-phase hydrogen cyanide; volatile organic, carbonyl, and phenolic compounds; and particulate-phase PAHs. Five replicates were analyzed for each of the three brands tested. Analyses were performed by Arista Laboratories (Richmond, Virginia). Tar, nicotine, and carbon monoxide concentrations were measured using the relevant ISO standards (3308; 8454; ). Nicotine and water were determined from iso-propanol extracts of total particulate matter retained by a collection pad using GC. Tar was calculated by subtracting nicotine and water values from the measured wet total particulate matter. Carbon monoxide was measured by nondispersive infrared spectroscopy. Mainstream smoke constituent yields for volatiles, phenolics, carbonyls, and PAHs were determined using high-pressure GC or GC with a mass selective detector. Hydrogen cyanide was determined using spectrophotometric analysis at 575 nm. Analytic methods are detailed in Arista Standard Operating Procedures (Arista Laboratories, Richmond, VA). Smoke chemistry data transformation Mainstream smoke constituents yields for MUS-SLC and MUS-Tampa were averaged across the five replicates for each brand. Constituent yields were compared with MUL, after controlling for nicotine yield of each brand. This manipulation provides a statistical control for the tendency for smokers to adjust smoke intake to accommodate low nicotine yield (Harris, 2004). Nicotine-controlled constituent yields of MUS were then expressed as percentages of the corresponding nicotine-controlled yield of MUL. Results Major design features of MUS prototypes The Salt Lake City prototype (MUS-SLC) uses a cavity filter design, in which 180 mg of synthetic

4 1200 NEW POTENTIAL REDUCED EXPOSURE PRODUCT carbon beads are located within a space between two cellulose acetate filter plugs. MUS-Tampa and MUS- Atlanta brands feature differing amounts of carbon granules derived from a natural source (120 mg and 45 mg, respectively) and distributed throughout a section of the cellulose acetate filter plug. MUL-ND is similar in design to MUS-Atlanta, with 45 mg of granular carbon in a carbon-on-tow filter. However, MUL-ND has lower filter ventilation (46%) than MUS-Atlanta (50%), and a lower total pressure drop (111 mmh 2 O vs. 115 mmh 2 O). The filters of MUS-Tampa and MUS-SLC are longer than the conventional ultra-light brand, reducing the total length of the tobacco column in these prototypes. Consequently, MUS contains less tobacco by weight than MUL. Analysis of filter carbon The carbon contained in the MUS-Atlanta and MUS-Tampa brands demonstrated a clear physical resemblance to carbon contained in other commercial charcoal-filtered cigarettes, including Parliament and Lark cigarettes. The carbon particles of these brands were consistent in structure and composition with activated charcoal. However, in the MUS- SLC prototype, the particles were dramatically different in appearance. The composition of these particles appeared to be a form of vitreous carbon, presenting a smooth and uniform surface and interior. Analysis of the cigarette filter components using SEM suggested that the MUS-SLC charcoal is a type of nonreactive, nonporous charcoal (vitreous carbon). SEM images of regular granular charcoal found in the other MUS prototypes reveal the structural contrast to MUS-SLC. The granular charcoal appears in SEM images as carbonized wood, in which the cellular structure of the wood fiber and the porous nature of the charcoal can be readily seen. The SEM images of MUS-SLC and MUS-Tampa charcoal are shown in Figure 1. SEM images of MUS-Atlanta appeared almost identical to MUS-Tampa and are not presented in the figure. Chemical analyses of artificial flavor additives A 6-mm fiber thread was located in the cellulose plug, below the carbon-containing space, of the filter of the MUS-SLC cigarette. Exploratory analyses for added artificial flavorants were conducted on this thread and on the carbon beads of MUS-SLC. The chromatograms from GC-MS analysis of both revealed high amounts of triacetin and propylene glycol. Degradants of the triacetin, namely monoand diacetin fragments, also were found in the extracts. Among the trace materials detected and tentatively identified by mass spectral library searches of the Wiley Mass Spectral Database were the flavorants 2-cyclohexen-1-ol, megastigmatrienone, menthol, and ethyl butyrate. The fragrance compound dihydrotrimethylphenylindene also was observed recurrently in the extraction solvents of the cotton fiber and charcoal. The presence of long-chain hydrocarbons was noted. Smoke chemistry Mean machine yields of mainstream MUS smoke constituents using FTC and HC regimens are reported in Tables 2 and 3. Mean raw yields for MUL are presented with MUS yield data, which are expressed as a percentage of MUL, after standardizing all brands for nicotine content. Based on the standard FTC regimen, yields of nicotine, tar, and carbon monoxide proved to be similar among the two MUS brands and MUL. Substantial reductions (.75%) in gas-phase volatile and carbonyl constituents, except formaldehyde, were observed under the FTC regimen. The median yields of six volatile-class constituents of MUS-SLC and MUS- Tampa were 14% and 6% of MUL, respectively, after standardizing for nicotine. Gas-phase carbonyl compounds of MUS-SLC and MUS-Tampa produced Table 1. Physical design features of Marlboro UltraSmooth and Marlboro Ultra Lights. Marlboro UltraSmooth Marlboro Ultra Lights Salt Lake City Tampa Atlanta North Dakota Conventional Tar yield (mg) a Tobacco weight (mg) Cigarette length (mm) Filter length (mm) Filter weight (g) Tipping length (mm) Tobacco column length (mm) Carbon type Bead Granular Granular Granular n/a Carbon weight (mg/cigarette) a Filter ventilation (%) Total pressure drop (mm H 2 O) Note. a Data reported by Philip Morris (2005).

5 NICOTINE & TOBACCO RESEARCH 1201 Figure 1. Scanning electron microscope images of Marlboro UltraSmooth (MUS) filter charcoal. Top left panel: MUS- SLC, whole bead; top right panel: MUS-SLC, bead section showing interior structure. Bottom left panel: MUS-Tampa, whole granule; bottom right panel: MUS-Tampa, granule interior. similar respective median yields of 14% and 6% of MUL. However, lesser reductions were observed on yields of gas-phase phenolic and particulate PAH compounds among the MUS brands compared with MUL. Phenolic compound yields for MUS-SLC and MUS-Tampa were similar to MUL (median yields of 90% and 83%, respectively). Median PAH compound yields of the MUS brands were equal to or greater than MUL (100% for MUS-SLC and 123% for MUS- Tampa). Hydrogen cyanide was substantially reduced in the MUS-Tampa brand (6% of MUL) and moderately reduced in the MUS-SLC brand (33% of MUL). As expected, constituent yields were higher under the intensive HC regimen than under the standard FTC method. Nicotine yields across all brands tested were approximately three times greater under the HC regimen. However, other constituent yields of the two MUS brands increased more substantially relative to MUL under the HC regimen when compared with the FTC regimen. As a result, reductions among volatile and phenolic constituents relative to MUL were of smaller magnitude under the HC regimen than under the FTC regimen. Median volatile compound yields were 27% (MUS-SLC) and 33% (MUS-Tampa) of MUL when adjusted

6 1202 NEW POTENTIAL REDUCED EXPOSURE PRODUCT Table 2. FTC method mainstream smoke constituent yields as nicotine-controlled percentage of Marlboro Ultra Lights. Marlboro Ultra Lights (MUL) Marlboro UltraSmooth (MUS) Salt Lake City Marlboro UltraSmooth MUS Tampa 0 mg carbon 180 mg carbon 120 mg carbon Nicotine (mg/cigarette) 0.56 (0.01) 0.53 (0.04) 0.42 (0.04) Constituent mg/cigarette MUS/MUL (%) Tar 6.3 (0.2) Carbon monoxide 8.6 (0.4) Volatiles mg/cigarette MUS/MUL (%) 1,3-butadiene 23.4 (2.0) b Isoprene (22.0) Acrylonitrile 5.6 (0.6) 24.7 b 9.6 a Benzene 31.3 (2.4) a Toluene 50.8 (4.5) a Styrene 3.5 (0.7) 24.4 b 7.7 a Median Phenolics mg/cigarette MUS/MUL (%) Hydroquinone 27.5 (1.7) Resorcinol 0.7 (0.03) Catechol 27.8 (1.3) Phenol 5.3 (0.6) m+p-cresol 4.5 (0.4) o-cresol 1.5 (0.2) Median Carbonyls mg/cigarette MUS/MUL (%) Formaldehyde 11.7 (2.5) Acetaldehyde (61) Acetone (22) b Acrolein 40.8 (5.9) b Propionaldehyde 36.7 (4.6) b Crotonaldehyde 14.1 (2.4) b Methylethylketone 56.0 (6.5) b Butyraldehyde 23.0 (2.6) b Median Polycyclic aromatic hydrocarbons ng/cigarette MUS/MUL (%) Naphthalene (22) b Fluorene (0.0) Phenanthrene (6.0) Anthracene 30.6 (0.7) Fluoranthene 38.0 (1.7) Pyrene 27.3 (1.3) Benzanthracene 8.5 (0.1) Chrysene 9.0 (0.3) Benzo(e)pyrene 2.4 (0.1) Benzo(a)pyrene 3.3 (0.2) Indeno[cd-1,2,3]pyrene 1.6 (0.1) Median Hydrogen cyanide mg/cigarette MUS/MUL (%) Hydrogen cyanide total 76.8 (5.9) Note. Values with parentheses are means with standard deviations. a Observed response at or below level of detection. b Observed response at or below level of quantification. for nicotine. Median carbonyl yields were 37% (MUS-SLC) and 57% (MUS-Tampa) of MUL. Phenolic (116% for MUS-SLC and 94% for MUS- Tampa) and PAH yields (85% and 96%, respectively) of both MUS brands were similar to MUL. Hydrogen cyanide was somewhat reduced in each MUS brand compared with MUL, yielding 62% (MUS-SLC) and 49% (MUS-Tampa) of the comparison brand. Discussion Among the four MUS prototypes, MUS-SLC comprises the greatest technological innovation, with vitreous carbon beads rather than granular charcoal, a discrete cavity housing the carbon rather than distribution through the cellulose filter tow, and an embedded cotton fiber purportedly for artificial flavor addition. These innovations appear intended

7 NICOTINE & TOBACCO RESEARCH 1203 Table 3. Health Canada method mainstream smoke constituent yields as nicotine-controlled percentage of MUL. Marlboro Ultra Lights (MUL) Marlboro UltraSmooth (MUS) Salt Lake City Marlboro UltraSmooth MUS Tampa 0 mg carbon 180 mg carbon 120 mg carbon Nicotine (mg/cigarette) 1.60 (0.12) 1.50 (0.06) 1.32 (0.11) Constituent mg/cigarette MUS/MUL (%) Tar 24.0 (2.3) Carbon monoxide 24.3 (1.8) Volatiles mg/cigarette MUS/MUL (%) 1,3-butadiene 78.2 (6.9) Isoprene (88.0) Acrylonitrile 26.0 (2.3) Benzene 85.4 (7.3) Toluene (10.0) Styrene 21.8 (2.1) Median Phenolics mg/cigarette MUS/MUL (%) Hydroquinone 68.9 (2.5) Resorcinol 2.1 (0.1) Catechol 61.8 (1.3) Phenol 11.2 (0.5) m+p-cresol 6.9 (0.3) o-cresol 2.5 (0.1) Median Carbonyls mg/cigarette MUS/MUL (%) Formaldehyde 63.2 (4.6) Acetaldehyde (168) Acetone (61.0) Acrolein (23.0) Propionaldehyde (14.0) Crotonaldehyde 54.2 (8.1) Methylethylketone 186 (21.0) Butyraldehyde 68.5 (8.7) Median Polycyclic aromatic hydrocarbons ng/cigarette MUS/MUL (%) Naphthalene (121) Fluorene (25.0) Phenanthrene (47.0) Anthracene 85.5 (11.8) Fluoranthene (14) Pyrene 77.7 (9.2) Benzanthracene 32.3 (5.8) Chrysene 32.4 (5.7) Benzo(e)pyrene 7.6 (1.1) Benzo(a)pyrene 12.5 (1.9) Indeno[cd-1,2,3]pyrene 5.9 (0.6) Median Hydrogen cyanide mg/cigarette MUS/MUL (%) Hydrogen cyanide total (51.0) Note. Values with parentheses are means with standard deviations. to support smoke constituent reduction while preserving or enhancing flavor characteristics. The present observations are consistent with the U.S. patent registered to Philip Morris that describes a MUS-like product (Jupe et al., 2004). Several structural differences between the MUS prototypes and the conventional MUL brand were observed. Although MUS cigarettes were similar in length to MUL, substantial differences were found in the relative lengths of the filters and tobacco columns of MUS-SLC and MUS-Tampa. The increase in filter lengths of these brands appears to be designed to accommodate their higher carbon loadings. Ventilation was higher among three MUS brands (SLC, Tampa, and Atlanta) compared with MUL. The highest ventilation observed was in MUS-SLC (58%), which contrasted with MUL (46%). This increased ventilation could contribute to the lower machine-generated smoke yields and also may partially explain the increased yields generated under

8 1204 NEW POTENTIAL REDUCED EXPOSURE PRODUCT the more intensive HC conditions relative to MUL. MUS-SLC and MUS-Tampa were found to have a lower pressure drop, or resistance to draw, than MUL, reducing the amount of work necessary to obtain a given smoke yield. The difference in the carbon used in MUS-SLC, in comparison with the other MUS prototypes and other commercial charcoal-filtered brands, was unexpected. As illustrated by SEM (Figure 1), the MUS-SLC filter contained spherical carbon beads of uniform size, with a smooth surface and low porosity. So-called vitreous carbon is appropriate for applications that require nonreactivity, extremely low porosity, thermal stability, high conductivity, chemical resistance, and smooth surface characteristics. Compared with vitreous carbon, natural charcoal fragments appear to have more surface area to react with smoke constituents and would be expected to provide a more effective means of carbon filtration. The reasons for Philip Morris s use of vitreous carbon in MUS-SLC are not clear. The spherical shape of the synthetic carbon beads could allow more efficient packing of carbon into a defined space. The MUS-SLC brand has achieved a 50% increase in carbon load compared with MUS- Tampa without increasing the length of its filter. The bead carbon may therefore have been used to make possible the placement of the relatively large 180-mg carbon load of MUS-SLC. The greater quantity of bead carbon could help to overcome its potential lesser capacity as an adsorbent of smoke constituents. Investigation of possible chemical flavor additives in the MUS-SLC brand revealed several potential candidates. GC-MS analysis of the flavor fiber found recurring evidence of ketones, aliphatic secondary alcohols, and related esters with known use as flavoring agents. For example, 2-cyclohexen-1-ol, a known component of spearmint, was detected in the methylene chloride extract. The methylene chloride extract of the flavor fiber also indicated the presence of dihydrotrimethylphenylindene, a known fragrance. Not unexpectedly, high amounts of the common filter additives triacetin (a plasticizing agent) and propylene glycol (a humectant) were found, which complicated the analysis of the flavor fiber. Efforts to partition the components into various solvents were marginally successful, and it is possible that other methods for sample preparation, such as solid-phase extraction techniques, are required to purify and concentrate the trace levels of flavorants and fragrances that appear to be present on the flavor fiber. Thus, although the flavor fiber as described in the Philip Morris patent was found and appeared to contain known flavoring agents and fragrances, improved or additional methods are needed to make definite conclusions regarding the presence and identity of any artificial flavorants in the flavor fiber of MUS-SLC. The carbon beads demonstrated more compounds under GC-MS analysis than the cotton fiber. Some of these compounds suggest absorption of semivolatile components. For example, components of tobacco, such as neophytadiene, nicotyrine, and nicotine, were detected in both the acetone and methylene chloride extracts of the charcoal. In addition, the known cigarette flavorants megastigmatrienone, menthol, and ethyl butyrate, as well as other possible flavor additives (e.g., heneicosane), were found in the charcoal extracts. Therefore, another possible role for the carbon beads is as a mechanism for flavor release. The data from the FTC smoking method suggest that the new MUS filter technology results in reductions of 75% or more in some toxic gas-phase volatile and carbonyl compounds compared with the control brand. Known toxins isoprene, benzene, toluene, acetone, acrolein, crotonaldehyde, and methylethylketone (International Agency for Research on Cancer, 2004), were reduced in the MUS-Tampa prototype to less than 5% of the MUL yield, under FTC conditions. The FTC method is the only smokeyield reporting requirement of the U.S. government. These data, taken alone, would imply that MUS could have important health outcome advantages compared with the conventional MUL. Such reductions in machine-derived constituent yields could provide some evidence of lower emissions and, potentially, lower exposure and reduced harm. Thus, in a superficial sense, MUS might attract consideration as a candidate product for reducing some forms of exposure. The MUS-Tampa prototype appeared to perform slightly better than MUS-SLC, with evidence of slighter greater reductions in volatile and carbonyl compounds compared with MUS-SLC. However, implications of the FTC-based reductions in yield of some MUS gas-phase constituents requires further scrutiny. A comparison between data obtained with the standard FTC and intensive HC regimens suggests that reductions in MUS yields compared with MUL depend heavily on the intensity of the smoking regimen. The intensive HC smoking protocol produced uniformly lesser reductions in MUS yield compared with control brand MUL when adjusting for nicotine delivery. For example, under the FTC method, 1,3-butadiene, a constituent with a high cancer risk index, was reduced by 88% in the MUS-Tampa prototype (Fowles & Dybing, 2003). This finding suggests virtual elimination from MUS of one of the most harmful tobacco smoke constituents, compared with the control MUL. However, under the HC method, the MUS-Tampa brand produced a reduction in 1,3-butadiene of only 19% compared with MUL.

9 NICOTINE & TOBACCO RESEARCH 1205 Indeed, the yields of all major gas-phase constituents, proportionate to MUL, were greater under the intensive HC regimen than under the FTC regimen, after adjusting for nicotine. Median FTC yields of MUS-SLC and MUS-Tampa volatile constituents were 14% and 6% of MUL, respectively, compared with 27% of MUL for MUS-SLC and 33% for MUS- Tampa, using the HC method. In the latter case, this represents a nearly sixfold greater increase in comparison with the increase in nicotine under the HC smoking regimen. MUS carbonyl compound yields increased to an even greater extent relative to MUL, from 14% to 37% (MUS-SLC) and from 6% to 57% (MUS-Tampa), representing 3-fold and 10- fold increases in compound yields, respectively, in comparison with the increase in nicotine under the HC smoking regimen. These simple comparisons between MUS and MUL yields suggest that although MUS may reduce some gas-phase smoke constituents, these reductions are lessened relative to nicotine delivery as a more intensive smoking protocol is applied. This finding suggests that the carbon technology of MUS is less capable of reducing gas-phase smoke constituents under more intensive smoking parameters. This may have important implications for the status of carbonfiltered low-yield PREP products, given that increasing intensity of actual human smoking behavior of this low-yield product might result in a decreasing efficacy of the carbon filter. Further research is required to explore this possibility. MUS performed poorly in reducing PAH compounds. This is not surprising, given that particulatephase constituents are known to be largely unaffected by carbon filtration (Hoffman & Hoffman, 1997; Philip Morris, 2002). Under the FTC method, median PAH yields were proportionately the same or higher, compared with yields from the conventional MUL, whereas yields under the HC method showed mild reductions of up to 13% of MUL. Unexpectedly, and perhaps more significant, there were no substantial reductions in gas-phase phenolic compounds. The median reduction in phenolic constituents of MUS-Tampa, when tested under the FTC method was 17% of MUL when adjusted for nicotine. Under the HC regimen, median phenolic yield for MUS-Tampa was less than 7% of MUL, whereas median phenolic yield of MUS-SLC exceeded the yield for MUL. Gas-phase carbon monoxide also was relatively unaffected by the carbon filters of MUS and showed yields comparable with MUL when adjusted for nicotine. The evidence assembled here suggests a concerted attempt by Philip Morris to produce a reduced exposure product while maintaining consumer appeal. Investigation of the product design shows that MUS has been carefully designed to hold high carbon loads while maintaining the physical appearance of a conventional cigarette. Cigarette ventilation and pressure drop have been manipulated to optimize smoker satisfaction while enhancing mainstream smoke dilution, and therefore, delivery of smoke constituents. An artificially enhanced flavor system is implicated in the design of MUS, as shown by discovery of the flavor fiber in MUS-SLC, and preliminary evidence regarding the presence of flavor additives was demonstrated. Finally, the smoke chemistry yield analyses suggest that although MUS may be a candidate for reducing some gasphase toxic constituent yields, and by implication reducing tobacco-related harm, its capacity to reduce toxic yields is less effective under a more intensive smoking regimen. Other highly toxic compounds including phenolics, hydrogen cyanide, and PAHs were reduced little if at all, compared with the lowyield conventional brand. One of the important limitations of the present study is the lack of smoke constituent yield data on the MUS-Atlanta and MUL-ND prototypes. Addition of the MUS-Atlanta and MUL-ND brands to the smoke chemistry analyses would have extended the range of carbon loads and filter designs investigated. Although it is presumed that the brands investigated, MUS-SLC and MUS-Tampa, should produce the greatest reductions in smoke constituents owing to their greater carbon loads, this presumption could be verified empirically with further testing of lower carbon load prototypes. Furthermore, it is unclear whether bead carbon provides greater constituentreduction capacity than granular carbon, as the prototypes tested contained unequal quantities (180 g bead compared with 120 g granule). Inclusion of the MUS-Atlanta (45 g granule) prototype may have provided further information regarding the efficiency of different granular carbon loadings under standard and high-intensity puffing conditions. Finally, caution should be exercised in making farreaching conclusions about the constituent-reduction capacity of MUS, based on a limited set of smoke constituents. In particular, tobacco-specific nitrosamine and heavy metal yield data were not obtained. These compounds are among the greatest estimated tobacco-related contributors to cancer risk (Fowles & Dybing, 2003). However, as particulate-phase constituents, nitrosamines and heavy metals are not likely to be reduced by the MUS carbon filter. The present study showed that the MUS prototypes tested demonstrated a capacity to reduce certain mainstream gas-phase smoke constituents, but this capacity was limited under more intense smoking conditions. MUS prototypes also had almost no effect on reducing certain particulatephase constituents. Further evaluation of MUS is required to draw conclusions regarding the

10 1206 NEW POTENTIAL REDUCED EXPOSURE PRODUCT likelihood of that this product will reduce exposure and tobacco-related harm. The effect of increasing machine-smoking intensity suggests that puffing parameters have an important influence on toxicant delivery of carbon-filtered PREPs. The carbon filter of MUS is less efficient at reducing mainstream smoke constituents under higher intensity puffing conditions. Studies of human smoking topography are required to determine the actual puffing parameters used by regular smokers when using MUS. Ideally, these parameters should be considered when using a machine-generated smoke-yield method to assess constituent yields in the future. Although the present study provides useful preliminary evidence to help guide evaluation of MUS and its potential use and effects, additional and longer-term studies will be required to perform a comprehensive assessment, including analysis of appropriate biomarkers for disease and disease risk. Research also is required to investigate the potential for false or ambiguous manufacturer health claims regarding PREPs such as MUS to influence the recruitment of new or ex-smokers or to delay cessation in current smokers. Misplaced consumer perceptions about the benefits of MUS and other PREPs may result in continued population health risks, just as light cigarettes were embraced by smokers for reasons of health and safety 20 years ago (Stratton et al., 2001). Clinical and epidemiological studies to monitor patterns of consumption and incidence of disease for MUS and other PREPs are critically needed. Acknowledgments This work was supported by an American Legacy Foundation, Evaluation and Research Initiative Grant (No. 6212). The authors are grateful to Heather Borski, Aaron Czyzewski, Celene Bardales Mulholland and Sunni Mumford for their assistance in obtaining cigarettes from test market cities. References Benowitz, N. L., Jacob, P., Kozlowski, L. T., & Yu, L. (1986). Influence of smoking fewer cigarettes on exposure to tar, nicotine, and carbon monoxide. The New England Journal of Medicine, 315, Djordjevic, M. V., Hoffmann, D., & Hoffmann, I. (1997). Nicotine regulates smoking patterns. Preventive Medicine, 26, Federal Trade Commission. (1980). Cigarettes and related matters: Carbon monoxide, tar and nicotine content of cigarette smoke; description of new machine methods to be used in testing. Federal Register (July 10). Fowles, J., & Dybing, E. (2003). Application of toxicological risk assessment principles to the chemical constituents of cigarette smoke. Tobacco Control, 12, Hammond, D., Fong, G. T., Cummings, K. M., O Connor, R. J., Giovino, G. A., & McNeill, A. (2006). Cigarette yields and human exposure: A comparison of alternative testing regimens. Cancer Epidemiology, Biomarkers, and Prevention, 15, Harris, J. E. (2004). Incomplete compensation does not imply reduced harm: Yields of 40 smoke toxicants per milligram nicotine in regular filter versus low-tar cigarettes in the 1999 Massachusetts Benchmark Study. Nicotine & Tobacco Research. 6, Erratum in Nicotine & Tobacco Research, 7, 173. Health Canada. (1999). Determination of tar, nicotine and carbon monoxide in mainstream tobacco smoke Official method. Ottawa: Author. Hoffmann, D., & Hoffmann, I. (1997). The changing cigarette, Journal of Toxicology and Environmental Health, 50, International Agency for Research on Cancer. (2004). IARC monographs on the evaluation of carcinogenic risks to humans, vol. 83: Tobacco smoke and involuntary smoking. Lyon, France: Author. Jupe, R., Dwyer, R. W., Laslie, D. E., Finley, A. L., Taylor, B. A., & Smith, C. M., et al. (2004). Cigarette and filter with downstream flavor addition. U.S. Patent No. 6,761,174. Washington, DC: U.S. Patent and Trademark Office. Kozlowski, L. T., O Connor, R. J., & Sweeney, C. T. (2001). Cigarette design. In, Risks associated with smoking cigarettes with low machine-measured yields of tar and nicotine. (Smoking and Tobacco Control Monograph No.13; NIH Publication No , pp.13 37). Bethesda, MD: U.S. Department of Health and Human Services. Philip Morris. (2000). SCoR 2002 product / process development. Bates No Retrieved from edu/cgi/getdoc?tid5hbo43a00&fmt5pdf&ref5results Philip Morris. (2001a). SCoR program objective. Bates No Retrieved from ucsf.edu/cgi/getdoc?tid5ktx94c00&fmt5pdf&ref5results Philip Morris. (2001b). Data from INBIFO report Bates No Retrieved from ucsf.edu/cgi/getdoc?tid5ybs41c00&fmt5pdf&ref5results Philip Morris. (2001c). Flavor review. Bates No Retrieved from getdoc?tid5wtf34a00&fmt5pdf&ref5results Philip Morris. (2002). Evaluation of activated carbon for use in cigarette filters. Bates No Retrieved from 5results Stratton, K., Shetty, P., Wallace, R., & Bondurant, S. (Eds.). (2001). Clearing the smoke: Assessing the science base for tobacco harm reduction. Washington, DC: National Academies Press.