Robustness of HPHC Reduction for THS 2.2 Aerosol Compared with 3R4F Reference Cigarette Smoke Under High Intensity Puffing Conditions

The reference cigarette used was 3R4F, supplied by the University of Kentucky (University of Kentucky Center for Tobacco Reference products, Lexington, KY, USA; http://www.ca.uky.edu/refcig/).

Commercially produced non-mentholated (regular) tobacco-sticks, known as HeatSticks or HEETS, designed for use with the THS 2.2 device were used. Published studies (7, 8) describe in detail the design and functionality principles for THS 2.2.

In one published study (9) an established list of principal aerosol/smoke constituents and 54 relevant HPHCs was used to assess the levels in THS 2.2 aerosol, compared to the mainstream smoke of the 3R4F reference cigarette. The same list of 54 relevant HPHCs (Supplementary information Table S 1) was considered in the present study.

The HCI regime is the machine-smoking regime commonly used as a reference, which was designed to generate emissions under a more intensive set of smoking parameters that would provide a ‘maximum’ exposure limit that could be exceeded by very few smokers. However it is widely recognized that even such an ‘intensive’ regime does not adequately represent the extremes relevant for human smoking behavior. Without specific standards and recommendations for testing robustness, alternative puffing regimes used to generate THS 2.2 aerosol were selected based on human puffing conditions observed with users (7), including an extreme regime representing very high-end usage conditions (110 mL puff volume, 4.5 s puff duration, 22 s puff interval). The most extreme regime used to generate 3R4F mainstream smoke (80 mL puff volume, 2.4 s puff duration, 25 s puff interval) was selected to exceed alternative high-end regimes used in previously reported cigarette studies (10, 11).

The HCI standard requires 100% occlusion of cigarette filter ventilation. For this investigation, both 100% and 50% occlusion of 3R4F filter ventilation was applied for each regime used, to evaluate the influence of this parameter on the robustness for HPHC reduction. Blocking ventilation was out of context for THS 2.2 because the tobacco-stick is designed without ventilation.

The analyses were performed at Labstat International ULC (Kitchener, Ontario, Canada), an ISO/IEC 17025:2005 (12) laboratory accredited for all the mandated tobacco-related Health Canada methods (13). The cigarettes and tobacco-sticks were conditioned according to ISO 3402:1999 (14). The test methods and method modifications for the analysis of THS 2.2 aerosol and 3R4F mainstream tobacco smoke are referenced in Supplementary information Table S 1 and were applied as described therein.

Five replicates were used for the analyses, unless specified otherwise.

Five different regimes described in Table 1 were used to generate THS 2.2 aerosol: augmented regime (A), extreme regime (E), high regime (H), intense regime (N) and low regime (L). These arbitrary names were defined for the identification of standard and non-standard regimes used for the aerosol generation of THS 2.2. Regimes L and N applied to THS 2.2 were representative of the puffing conditions for the ISO (ISO) and Canada Intense (HCI) standards, respectively. The THS 2.2 system is programmed to finish heating after a maximum period of 6 min or after 14 puffs, whichever comes first. Accordingly, the puff intervals applied for the ISO (60 s) and HCI regimes (30 s) resulted in the generation of 6 puffs and 12 puffs, respectively.

Three different regimes described in Table 2 were used to generate 3R4F mainstream smoke. For each regime, 3R4F reference cigarettes were smoked with 50% and 100% ventilation occlusion, achieved by applying adhesive tape strips. Irrespective of the level of ventilation occlusion, the low regime (L) and the intense regime (N) represent puffing conditions for the ISO (ISO) and Health Canada Intense (HCI) standard regimes, respectively.

The HPHC aerosol/smoke concentrations were calculated as mass per tobacco-stick normalized per total puff volume for each product, constituent, regime and replicate. The number of puffs multiplied by the volume of puffs determined the total puff volume. The number of puffs averaged between all items of a replicate was used for the 3R4F cigarette and the number of puffs predefined by the regime was used for the THS 2.2. These data were used for the entire evaluation process as described below. The mean and standard deviation of yields and volume-adjusted yields are reported in Table 6 and Table 7, respectively.

Nicotine-adjusted HPHC yields were calculated as mass per tobacco-stick normalized for the corresponding nicotine yield for each product, constituent, regime and replicate. These data were used for the calculation of average reductions for THS 2.2 versus 3R4F per regime, using data for all HPHCs combined. The mean and standard deviation of nicotine-adjusted yields are reported in Table 8.

Descriptive statistics per product and machine-smoking condition were computed for all HPHCs investigated in this study.

The relative reduction was defined as the reduction of the HPHC concentration in THS 2.2 aerosol compared with that in 3R4F cigarette smoke.

Statistical analysis Bayesian ANOVA (15, 16) was performed to assess the reduction of HPHCs across the different machine-smoking conditions. The mean concentrations and respective 0.95 confidence intervals for each HPHC were calculated for each of the five conditions tested for 3R4F and each of the six conditions tested for THS 2.2. Relative reductions for the levels of individual HPHCs and their respective 0.95 confidence intervals were calculated for the 30 combinations of conditions using Markov Chains produced by the Bayesian model (15, 17). For each of the 30 combinations, the average reductions were calculated for both nicotine-adjusted and volume-normalized yields using combined data from all the HPHCs measured. Pre-defined standard distributions, i.e. non-informative priors, were assigned to all Bayesian model parameters to match the classical frequentist modelling results. The Bayesian approach allowed the computation of uncertainty around complex statistical estimates.

SAS Enterprise Guide 7.1 was used for the statistical treatments, including Procedures MCMC and MIXED (18, 19) for the computation aspects.

Aerosol and smoke constituent values were estimated when they were below the limit of detection (LOD) and/or lower limit of quantitation (LOQ) of the analytical method. For the THS 2.2, the values used to calculate the relative reduction between THS 2.2 and 3R4F were imputed as follows: for results < LOD, the value was estimated by the LOD; for results < LOQ but > LOD, the value was estimated by the LOQ.

When more than two out of five replicate values used in the calculation of reduction were below the LOQ, the 0.95 confidence intervals were not reported. The number of acceptable values below the LOQ were considered individually in THS 2.2 and 3R4F emissions.

No HPHC reduction was reported for conditions presenting more than two out of five replicate values for 3R4F reference cigarette emissions below the LOQ.

Each HPHC was categorized to one of 11 individual threshold limits of robust reduction. The first threshold limit category was established for cases demonstrating a minimum of 50% reduction, and the next successive limits follow a progression using 5% increments until the last threshold limit being fixed at 99%.

The acceptance criteria in a category is the non-inferiority of the lower band of the 0.95 confidence interval of the HPHC reduction with the respective threshold limit. The test is unilateral and can detect non-inferiority with at least 0.975 probability. For each investigated HPHC, the highest threshold limit demonstrating non-inferiority of reduction for all regimes served to determine the threshold limit of robust reduction.

Testing 3R4F cigarettes with 50% occlusion of filter ventilation holes was performed to embrace a broader range of possible conditions and showed the pertinence of using puff volume for the normalization of yields, focusing not only on the THS 2.2 product, but also on the 3R4F cigarette. Interestingly, the results showed consistent overall reductions across all puffing intensities between the two different conditions of filter ventilation used for the 3R4F cigarette. However, in the scope of the assessment for THS 2.2 robustness, the shift in relative reductions caused by the change of 3R4F filter ventilation for some individual constituents presented less interest than the overall influence of puffing intensity. Therefore, unless stated otherwise, the following discussion has focused upon reductions calculated using HPHC concentrations for 3R4F cigarette mainstream smoke generated with 100% occlusion of filter ventilation. This is also consistent with the requirements of the Health Canada Intense (HCI) machine-smoking regime of the Health Canada Tobacco Reporting Regulation (22) that was recommended by Public Health authorities, including the World Health Organization (WHO) (23).

Using the Health Canada Intense regime, and considering 49 constituents, the average reduction for puff volume-adjusted HPHC concentrations in THS 2.2 aerosol compared to 3R4F cigarette smoke was 93.6% (88.7% using nicotine-adjusted values). This is in line with published results (8) where the majority of HPHCs measured in THS 2.2 aerosol were reduced by more than 90%. When more intense machine-smoking regimes were applied, average reductions based upon puff volume-adjusted concentrations for THS 2.2 were in the range from 94.2% to 95.9% (88.4% to 89.3% using nicotine-adjusted values), irrespective of the smoking regime applied for the 3R4F reference cigarette, demonstrating a very robust performance for overall HPHC reduction.

Each puff taken from a cigarette burns a new portion of the tobacco rod as opposed to the different zones of the unique portion of tobacco that are progressively heated and repeatedly used over puffs in THS 2.2. Therefore, depending on the nature of precursors and the release mechanisms of individual HPHCs, different levels of reductions are expected between individual HPHCs and regimes.

Metals and metalloids are absorbed in growing tobacco leaves at harvesting point (24) and some of them can be transferred in THS 2.2 aerosol. The threshold limit of robust reduction of 50% observed for mercury contrasted with arsenic, lead and cadmium having higher limits of 90%, 95% and 99%, respectively. The fact that mercury can distill out of tobacco at a relatively low temperature (25, 26) is a possible cause of this difference.

Heating a nitrogen-rich biomass such as tobacco produces ammonia (25). Acrylamide can be formed from asparagine and reducing sugars through a Maillard-type reaction occurring between 120 °C and 200 °C (27, 28, 29, 30). Acetamide can be formed by the pyrolysis of Amadori compounds, which are reaction products needing amino acids and sugars, and by the decomposition of ammonium acetate at around 250 °C (31). Pyridine can be formed below 300 °C through a classic Maillard reaction (32). These nitrogen-containing products are formed and distill out of tobacco at a low temperature. It was therefore unsurprising to observe threshold limits of robust reduction in the range of 55–75%. In contrast, the five investigated aromatic amines demonstrated a limit of 99%, which denoted consistently low levels of generation at THS 2.2 operating temperatures. Quinic acid and quinic acid derivatives, that are building blocks in the natural biosynthesis of lignin, were identified as important precursors of hydroquinone, catechol, and phenol which can be formed below 350 °C (33). Phenol concentration in THS 2.2 aerosol increased simultaneously with puffing regime intensity, which resulted in a diminution of phenol reduction from 92.2 ± 1.4% (HCI) to 63.2 ± 5.8% (extreme regime) for the lowest performance and was reflected by a threshold limit of reduction of 55%. This was possibly caused by the progressive saturation and the subsequent elution of phenol adsorbed on tobacco and other filters (34, 35).

PAHs can enter tobacco leaves during curing if the tobacco is exposed to exhaust gases from wood, or other organic fuel, heat sources (36). When studying the aerosol content of benzo[a]pyrene, benz[a]anthracene and pyrene in aerosols produced by different tobacco blends in the THS 2.2 (see Supplementary information Table S 9), it was observed that the yields of these PAHs were linked to the tobacco composition of the different blends (37). The results presented in Supplementary information Table S 9 show that the transfer of benzo[a]pyrene, benz[a]anthracene and pyrene in THS 2.2 was proportional to the PAH content of the tobacco materials of the sticks, and very good correlations (r > 0.99) were observed for these three PAHs when comparing yields in tobacco and in aerosol produced under HCI. Considering more intense regimes and supported by a threshold limit of robust reduction of 90% for benz[a]anthracene and pyrene and 95% for benzo[a]pyrene, the results showed that the PAHs detected in the aerosols of the THS 2.2 are most likely due to stripping caused by the gas stream and not to pyrosynthetic formation occurring during the THS 2.2 heating process. The largest quantities of TSNAs are formed during tobacco curing and processing (38, 39, 40) and additional amounts could be formed during smoking. TSNAs distillate out of tobacco at low temperature, nevertheless heating instead of burning tobacco in THS 2.2 results in a 2–3 times lower transfer of TSNAs when compared with the transfer rates found for cigarettes (41). Under HCI and more intense regimes, the limit of threshold reduction was 95% for NNN, NNK and NAT whereas the limit was 90% for NAB; this difference could be attributed to individual transfer rates of TSNAs (41, 42).

Carbon monoxide, HCN, nitrogen oxides and volatile organic compounds such as 1,3-butadiene, acrylonitrile, benzene, styrene and toluene are generated only at low levels in THS 2.2 aerosol during thermochemical degradation of the tobacco material and have demonstrated threshold limits of a robust reduction of 90% or higher. A slightly lower limit of 85% was observed for propylene oxide. The dehydration of propylene glycol used as humectant or for the application of flavors to tobacco is a possible source of propylene oxide in mainstream cigarette smoke (43, 44), however the absence of propylene glycol in the recipe of the 3R4F cigarette (45) suggests other possible precursors may have a role in the formation of propylene oxide. Amongst compounds with less clear sources, the limit was 95% for both vinyl chloride and ethylene oxide.

The threshold limits of robust reduction were in the range of 75–95% for all carbonyls investigated under HCI and more intense regimes. The lowest performance of reduction was for butyraldehyde and could denote a low temperature of formation. The threshold limit of robust reduction of 75% is in good agreement with values reported in the literature for tests conducted under HCI regimes, e.g., 71.13% (46) and 75.6% (47). Like most other carbonyls, butyraldehyde content in THS 2.2 aerosol steadily dropped with increasing puffing intensity, therefore the highest concentrations were observed in THS 2.2 aerosol generated with the lowest puffing intensities. For butyraldehyde under ISO conditions, a relatively low reduction of 41.92% was published (46). This is consistent with the reductions of 56 ± 3% and 43 ± 5% obtained in the present study with the same smoking-machine parameters, taking also into account a difference in the occlusion levels used for 3R4F filter ventilation, respectively 100% and 50%.