Puffing Topography, Mouth Level Exposure, Average Daily Consumption and Machine Emissions of Three Variants of glo™ Heated Tobacco Products Among UK Smokers

Three king size super slim (KSSS) HTP consumables (Kent Neostiks™; BAT, London, UK) were used with the same glo™ HTP device (Nicoventures, London, UK). The glo™ device (8) is a battery-powered electronic heating device that uses a resistive heating mechanism to produce an inhalable aerosol by heating the tobacco consumable, a paper rod containing reconstituted blended tobacco, to a maximum temperature of approximately 240 ± 5 °C (Figure 1). The device provides the consumer with a fixed 3.5-min heating session. The glo™ device, when used with the Bright Tobacco consumable containing 0.85% nicotine on a dry-weight basis (dwb) (base HTP), has reduced emissions, toxicological effects and toxicant exposure relative to a conventional cigarette (6).

Figure 1.

Diagram of glo™ HTP with a tobacco consumable.

Here, the base HTP consumable was re-evaluated alongside two variant consumables that differed either in tobacco flavour (Rich Tobacco, 0.85% nicotine dwb; Variant 1) or in tobacco flavour and nicotine strength (Rich Tobacco, 1.5% nicotine dwb; Variant 2). The base consumable and Variant 2 were commercially available in Japan and European markets at the time of the study; Variant 1 was not commercially available at the time of the study. None of the study products were commercially available in the UK at the time of the study. All devices and consumables used by study participants were stored at ambient indoor room conditions in a locked location for the duration of the study when not in use. Temperature and relative humidity were not actively measured or controlled, as this was intended to reflect the uncontrolled environmental conditions typical of actual human use.

Machine emissions analysis was conducted by Labstat (Kitchener, ON, Canada) to determine the quantity of key analytes in the aerosol of the three study products and the percentage reduction of nine toxicants (TobReg9) identified by public health authorities for potential reduction in cigarette smoke, relative to smoke from a certified 3R4F reference cigarette (Center for Tobacco Reference Products, University of Kentucky, Lexington, KY, USA). Before testing, all study HTP consumables and reference cigarettes were environmentally conditioned as specified in ISO 3402 (31).

HTP aerosol was generated using the ISO 20778:2018 machine-puffing regime of puff volume 55 ± 0.5 mL, puff duration 2.0 ± 0.1 s, puff frequency 30 ± 1 s, bell-shaped profile and 0% blocking of the perforations (26, 27). Five replicate consumables were tested per study product, with eight puffs taken per consumable (determined by the fixed 3.5-min heating session of the glo™ device, and the 30-s frequency specified in the machine-puffing regime).

The 3R4F reference cigarette was machine-smoked using the Health Canada Intense (HCI) regime without modification. glo™ aerosol and cigarette smoke were collected by certified and established methods set out in accordance with the International Organization for Standardization ISO 3308:12 (32).

The primary objective was to demonstrate statistical equivalence within predefined margins, between the base consumable (Bright Tobacco, 0.85% nicotine) and Variant 1 (Rich Tobacco, 0.85% nicotine) or Variant 2 (Rich Tobacco, 1.5% nicotine), by comparing MLE to NFDPM and nicotine (per stick and per day), and ADC among the three consumables. The secondary objective was to compare puffing topography, including puff volume, puff duration, inter-puff interval and puff number between the three consumables.

Analyses were carried out using SAS v. 9.4 (SAS Institute, Cary, NC, USA). All 72 participants completed the study and were included in the analysis. Two participants had equipment errors during topography sessions, therefore topography and MLE data are reported for n = 70 participants, and ADC data are reported for n = 72 participants. Puffing topography, ADC and MLE are reported as mean ± standard deviation (SD) per product. Puffing topography and MLE data were first averaged per participant across the two repeats. ADC was derived from the self-reported daily consumption diaries, and was calculated in two ways (i) ADC of the study product alone, and (ii) the total ADC, which was calculated as the sum of study product consumption plus non-study tobacco product consumption. Participants who self-reported zero study product or non-study tobacco product use contributed valid zero values and were therefore included in the calculations of ADC. Per day consumption was first averaged per participant from days 2–4 of the home use period. Days 1 and 5 did not represent full stable use days (device initiation and return days, respectively) and therefore were excluded from ADC calculations. Prior to statistical analysis, analytical thresholds were defined for each attribute based on the limitations of the methodology (Supplementary Figure S1).

These thresholds reflect the performance characteristics of the puffing topography equipment, including measurement resolution, repeatability, calibration uncertainty and inherent variability observed during method validation.

Differences smaller than these thresholds therefore cannot be reliably distinguished from measurement variability. For a given attribute, if a difference between two product measurements was less than the defined threshold, it was deemed “unmeasurable” within the limits of the analytical method, and therefore any statistical differences were deemed insignificant. Where the difference between measurements was greater than the defined analytical threshold, a paired two-one-sided t-test (TOST) procedure on the ratio of means, with lower and upper acceptance limits of 0.80 and 1.25 and 90% confidence intervals (CI), was used to assess equivalence to the base product. To our knowledge, no other published studies have applied equivalence margins or TOST to puffing topography endpoints. We therefore adopted the European Medicines Agency guidelines of 0.80–1.25 with 90% CI, which is the regulatory standard for pharmacokinetics (PK) based equivalence in orally inhaled products (33). Since puffing behaviour and MLE contribute directly to aerosol delivery and exposure, we considered this equivalence margin appropriate for this study. In addition to the primary TOST analysis, a sensitivity analysis was conducted using a linear mixed-effects model with fixed effects for product, period and sequence, and a random effect for subject. This model was used solely to assess whether period or sequence effects influenced the product comparisons. The TOST remained the basis for equivalence decisions.

Machine emissions data are reported as a percentage reduction relative to 3R4F reference cigarette smoke. For each TobReg9 toxicant, the percentage reduction was calculated per replicate as follows: % Reduction =100−(HTP/3R4F)×100 \% \,{\rm{Reduction}}\, = 100 - ({\rm{HTP}}/3{\rm{R}}4{\rm{F}}) \times 100

The mean of the replicates was then determined. Where toxicants were reported as below detection level, the value was reported as half of the level of detection (LOD). Where analytes were reported as not quantifiable, the value was reported as the midpoint between the LOD and level of quantification.

In total, 72 smokers were recruited to participate in the study, all of whom completed the home use and central location tests for all three study products. Puffing topography and MLE data are reported for n = 70 participants due to equipment error for two participants, and ADC data are reported for n = 72 participants. Among the participants, 36 (50%) were male, 36 (50%) were female, and the age range was 21–64 years. Participant flow is shown in Supplementary Figure S2.

The puffing topography attributes (number of puffs, puff volume, puff duration, inter-puff interval and effort expended) of the three products are summarised in Table 2. For Variant 1, which differed from the base consumable only in tobacco flavour, differences in puff number (16.4 vs. 15.9), puff duration (2.0 vs. 2.0 s) and puff interval (9.7 vs. 10.0 s) were smaller than the predefined analytical thresholds (Supplementary Figure S1). These differences were therefore considered not measurable, and equivalence testing was not applied. Differences in puff volume (63.4 vs. 58.9 mL; mean ratio 90% CI: 1.01, 1.12) and effort (526 vs. 500 mm WGs (millimetre water gauge seconds); mean ratio 90% CI: 0.99, 1.15) were within the predefined equivalence margin of 0.80–1.25 and were considered equivalent. For Variant 2 which differed from the base consumable in tobacco flavour and higher nicotine strength, differences in puff volume (58.8 vs. 58.9 mL) and puff interval (10.0 vs. 10.0) were smaller than the predefined analytical thresholds (Supplementary Figure S1), and therefore equivalence testing was not applied. Differences in puff number (14.8 vs. 15.9; mean ratio 90% CI: 0.86, 0.98), puff duration (1.8 vs. 2.0 s; mean ratio 90% CI: 0.86, 0.93) were within the predefined equivalence margin and were therefore considered equivalent, while the difference in effort (437 vs. 500 mmWGs; mean ratio 90% CI: 0.80, 0.94) fell outside the equivalence margin. Mixed effects modelling showed no significant period or sequence effects across all endpoints, indicating that consumer behaviour was not influenced by product order or study week.

The mean puff volumes and durations for all products were consistent with the 55-mL puff volume and 2-s puff duration of the ISO 20778:2018 machine puffing regime used to generate emissions data (26, 27). The mean inter-puff intervals (9.7–10.0 s) were less than the 30-s frequency specified in ISO 20778:2018, indicating that participants took puffs more frequently than the machine puffing regime. Within the fixed 3.5-min heating session of the device, these shorter puff intervals allowed participants to take more puffs per session (14.8–16.4) compared with the eight puffs obtained during machine emissions testing. Consequently, emissions generated using ISO 20778:2018 may underestimate actual consumer exposure. The inter-puff intervals here were consistent with those of previous studies using the same glo™ device (7.4–11.1 s) (23,24,25), supporting that participant puffing behaviour was representative of typical glo™ use.

Our observations showed that the puffing measures were very similar across the study products despite the differences in tobacco flavour (Variant 1), and tobacco flavour and nicotine content (Variant 2) as compared with the base consumable. This finding implies that any nicotine-induced sensations initiated during the inhalation phase did not have a controlling influence on participant puffing behaviour. This is consistent with the results of Buzzi et al. (34) and Dixon et al. (35), who reported that the nicotine yield of a cigarette did not influence puff-by-puff changes in volume and duration. Additionally, Dixon et al. (35) demonstrated that puff volume and duration were influenced by sensory responses occurring during puffing process (e.g., mouthful), but not by inhalation sensations (35).

Two previous studies have measured puffing topography attributes of the glo™ HTP and base consumable via a similar approach (23, 24). The current puff duration (2.0 s) was similar to previous values (1.6–1.8 s); puff volume (58.9 mL) and puff interval (10.0 s) fell within the previously reported ranges (46.6–66.7 mL and 7.4–11.1 s, respectively); and puff number was slightly higher in the present study (15.9 vs. 10.9–15.4) (23, 24). These differences may be due to variations in the study populations: the study of Gee et al. (23) was conducted in Japan among both smokers and regular HTP users, that of Jones et al. (24) was conducted among smokers in Italy, while the present study was conducted among smokers in the UK. More broadly, studies of HTPs by Prasad et al. (25) and Li et al. (36) reported higher puff numbers of 17–21 puffs and 22 puffs, respectively, while puff volumes, durations and intervals were generally within previously observed ranges [57–81 mL volume, 1.6–1.9 s duration and 9 s interval (25); and 59–66 mL volume, 1.9–2.0 s duration and 10 s interval (36)]. These cross-study comparisons are descriptive only; formal statistical testing between studies was not conducted due to differences in study populations (smoker/HTP-user mix), geographical locations (Japan, Italy, UK) and study design.

The consumables used in the present study were KSSS sticks (diameter ~5.0 mm; tobacco weight ~260 mg). Prasad et al. (25) reported that a larger-diameter demi-slim (DS) consumable (diameter ~6.8 mm; tobacco weight ~340 mg) produced notably lower open pressure drop (~45 mmWG vs. ~70 mmWG for KSSS) and correspondingly larger puff volumes (69–81 mL vs. 57 mL for KSSS) compared with a KSSS consumable, illustrating that airflow resistance driven by consumable geometry directly influences puff volume. In the same study, MLE to NFDPM was also higher for the DS consumable (18.8–19.9 mg/stick vs. 12.5 mg/stick for KSSS), likely reflecting a combination of the larger puff volumes and greater tobacco weight, although it is worth noting that the device paired with the KSSS consumable employed resistive heating, whereas the device paired with the DS consumable employed induction heating. Consumable flavour and nicotine content were chosen as variables of interest in this study since they represent common sources of product variation that may influence sensory perception characteristics (e.g., taste, impact) and nicotine delivery, which in turn could affect puffing behaviour and exposure. Evaluating these attributes can help determine whether such product variations remain behaviourally comparable to the established base consumable.

The puffing behaviour observed in the current study falls within the established range for KSSS-format glo™ consumables and is consistent with values reported for similar products. This supports the interpretation that participants' puffing behaviour in this study reflects typical use of glo™ consumables.

The self-reported study product ADC, total ADC (sum of study product ADC plus cigarette ADC), and MLE to NFDPM and nicotine (per stick and per day) of the three products are summarised in Table 3.

The number of participants reporting continued use of their usual brand cigarettes was 70 for the base consumable, 68 for Variant 1 and 69 for Variant 2. For Variant 1, differences in study product ADC (4.6 vs. 4.5 sticks/day), total ADC (12.7 vs. 12.6 sticks/day), MLE per stick (NFDPM: 5.3 vs. 4.7 mg; nicotine: 0.30 vs. 0.30 mg) and MLE to nicotine per day (1.56 vs. 1.40 mg) were smaller than the predefined analytical thresholds (Supplementary Figure S1), and therefore equivalence testing was not applied. Differences in MLE to NFDPM per day (27.2 vs. 22.2 mg; mean ratio 90% CI: 0.93, 1.23) fell within the equivalence margin.

For Variant 2, differences in study product ADC (4.6 vs. 4.5 sticks/day), total ADC (12.0 vs. 12.6 sticks/day), and MLE to NFDPM (per stick: 4.6 vs. 4.7 mg; per day: 22.4 vs. 22.2 mg) were smaller than the predefined analytical thresholds and therefore equivalence testing was not applied. MLE to nicotine was higher for Variant 2 than for the base consumable (MLE per stick: 0.41 vs. 0.30 mg; mean ratio 90% CI: 1.05, 1.34; MLE per day: 2.00 vs. 1.40 mg; mean ratio 90% CI: 1.03, 1.48), and did not meet the equivalence margin, and therefore equivalence could not be confirmed. This is consistent with the higher amount of nicotine in the tobacco blend for Variant 2 (1.50% vs. 0.85%). The MLE estimates reported within this study are consistent with previous topography studies conducted in Japan and Italy (23, 24) using the base product (Table 4), which is again consistent with similar overall puffing behaviour (Table 2).

The ADC reported for the base product among the UK smokers in the present study (4.5 ± 3.8 sticks/day) was much smaller than previously reported in Japan and Italy (23, 24). After home placement, the ADC of the base product in Japan was 10.3 ± 5.5 sticks/day among 52 smokers and 8.6 ± 4.6 sticks/day among 52 HTP users (23), and in Italy it was 7.0 ± 5.5 sticks/day among 52 smokers (24). As well as differences in the study populations, these variations may be due to the prevalence of HTP use, which is generally lower in the UK than in Japan and Italy (37, 38). Furthermore, unlike in Japan and Italy, glo™ HTPs were not commercially available in the UK at the time of the study. Taken together, this may explain why 70 of the 72 UK smokers in the current study continued to use their usual brand cigarette in addition to the study product at home. In support of this, the total ADC of the tobacco and nicotine products (Table 3) is fairly consistent across the three studies: 12.6 ± 6.0 sticks/day in the present study; 12.1 ± 5.5 sticks/day among smokers and 11.2 ± 6.2 sticks/day among HTP users in Japan (23); and 11.6 ± 5.7 sticks/day in Italy (24) (Table 4).

In summary, most observed differences between the base consumable and the variant consumables in puffing topography, MLE and ADC were either (i) not measurable according to the predefined analytical thresholds (Supplementary Table S1), or (ii) were within the predefined equivalence margins according to a paired TOST (90% CI of mean ratio). The only exceptions were MLE to nicotine (per stick and per day) for Variant 2, which was higher than the base consumable and consistent with higher nicotine content, and effort for Variant 2, indicating that differences in tobacco flavour and nicotine strength among the study products had no significant effects on user behaviour or product consumption.

The nine TobReg9 toxicants identified by public health authorities for potential reduction in cigarette smoke were significantly reduced in machine-generated emissions from all three study products relative to the 3R4F reference cigarette (90.61–99.99% reduction), with no differences observed in overall percentage reductions across the three products (97.02–97.93% reduction) (Table 5). Some small differences were observed in the percentage reduction of formaldehyde and N-nitrosonornicotine (NNN), where the percent reduction was greater for Variant 1 and Variant 2 than for the base product (95.60–96.32% vs. 93.92% reduction for formaldehyde and 93.72–93.74% vs. 90.61% reduction for NNN). No formal between-variant significance testing was performed for machine emissions data since the primary comparisons were to 3R4F cigarette smoke and the study was not powered for small between-variant emissions differences. Small differences between consumables are expected due to variations within batches of tobacco. We also compared machine emissions of nicotine, NFDPM, water and glycerol among the study products (Table 6).

The nicotine delivery for Variant 2 was higher than that of the base consumable and Variant 1 (0.55 vs. 0.38–0.46 mg/stick), consistent with the higher nicotine content in the blend. Although aerosol NFDPM was slightly higher for the base product relative to variants 1 and 2 (base product: 13.5 mg/stick; Variant 1: 11.0 mg/stick; Variant 2: 11.1 mg/stick), there were no differences between the variant consumables.

Despite participants puffing more frequently than the 30-s puff duration of the ISO 20778:2018 machine puffing regime, the mean of the estimated MLE to NFDPM and nicotine per stick (Table 3) remained lower than the corresponding machine-generated emissions (Table 6). This indicates that machine emissions represent the maximum potential aerosol yields under standardised conditions, whereas actual consumer exposure may be further influenced by other factors.