An Experimental Analytical and In Vitro Approach to Bridge Between Different Heated Tobacco Product Variants

The five variant THPs, THP_base, commercial MRTP-approved comparator THP (hereafter THP_benchmark/THS2.2_R), and reference cigarette assessed in the study are summarized in Table 1. The modifications of the variant THPs were classified as minor (small changes in either top flavour, nicotine strength, or aesthetics) or major (a significant change brought about by a combination of minor changes). Unless otherwise stated, all materials and reagents were purchased from Fisher Scientific (Loughborough, UK).

For the bridging dataset, we evaluated emissions and cytotoxicity. Emissions measurements were conducted on the THP_base and five variants for toxicants identified by the World Health Organisation Study Group on Tobacco Harm Reduction (TobReg9). (23). Historical data were compiled for cigarette smoke (16) and the THP_benchmark as a commercial comparator (7). All emissions analyses were conducted at an independent accredited laboratory, Labstat International ULC (Kitchener, ON, Canada) following the TobReg9 recommendations (23). Five replicates were performed per analysis. All analytical approaches, together with the limits of detection and quantification for each analyte, have been previously documented (16).

For cytotoxicity testing, human bronchial epithelial cells (NCI-H292; American Type Culture Collection, Teddington, Middlesex, UK) were grown and maintained in sterile-filtered RPMI-1640 medium for cell culture. The medium was supplemented with 10% foetal bovine serum (GE Healthcare Life Sciences, Hatfield, Hertfordshire, UK), 2 mM glutamine, 50 U/mL penicillin and 50 mg/mL streptomycin. Twelve-well cell culture plates containing Transwells^TM were seeded with 500 µL of cell suspension (~ 2×10⁵ cells). Cultures were incubated for 72 h at 37 °C in humidified 5% carbon dioxide before experimental exposure. Cells were exposed at the air-liquid interface to THP whole aerosol (WA) generated by a Borgwaldt LM4E vaping machine or 1R6F reference cigarette whole smoke (WS) generated by a Borgwaldt RM20D smoking machine as previously described (24, 25).

A full cytotoxicity dataset was generated over 1–180 puffs of undiluted WA from the THPs (THP_base, n = 9; variant THPs, n = 3) or 0–8 puffs of undiluted 1R6F WS (n = 3), which evoked a maximum cytotoxic response. On each exposure day, we carried out a sham air control (blank) in which filtered laboratory air was puffed onto the cells; the number of puffs corresponded to the highest puff number used on that day. After exposure, all chambers including the sham air control were left for 5–10 min to allow any aerosol to settle before the cells were processed to recovery plates for 24 h. A set of negative controls were also carried out on each exposure day: 3 media-submerged inserts (incubator controls); and 3 inserts at the air-liquid interface (ALI) prepared by removing the apical media and replacing the chamber back in the incubator (ALI controls). Treatment of cells with 350 mM sodium dodecyl sulphate (SDS) served as a positive control.

At 24 h after exposure, neutral red uptake (NRU) assay (15, 26) was used to assess cell viability. Cells were washed twice with sterile PBS and incubated with neutral red dye for 3 h, and then washed twice with PBS to remove excess dye. Uptake of dye by viable cells was determined by lysing the cells with 500 µL of de-stain solution and quantifying 100 µL aliquots on a microplate spectrophotometer at 540 nm using a reference filter of 630 nm. Cell viability was reported as a percentage of the air control.

Data were processed in GraphPad Prism 8 or Microsoft Excel. All cytotoxicity data were normalised to the air control response. Emissions data were compared using GraphPad Prism (version 8) and percentage reductions were compared across all products. Total analyte yields (TAY), which is the accumulation of all measured analytes, were calculated in Microsoft Excel. TAY was based on a per-puff basis for cigarette smoke and THP. Cigarette smoke TAY was based on 3 puffs (10.7 × 3 = 32.1 mg); THP TAY was based on 85 puffs (mg/puff × 85). Biological responses to undiluted WA were compared by one-way ANOVA (p < 0.05 was considered significant) of the generated IC₅₀s. Three cell samples were used per treatment condition. To compare the THPs, a single 85-puff dose was selected from the THP baseline data; this dose represented the first dose beyond the half-maximal inhibitory concentration (IC₅₀) and ensured that comparisons were made for an active dose. Cigarette smoke data were used to contextualise the THP responses.

THP emissions were measured for 12 analytes (Table 2). In all cases, levels in the five variant THPs were comparable to the THP_base. Relative to cigarette smoke, percentage reductions were between 94% and 97% for all six THPs (Table 2 and Figure 1a). All six products were compared by a one-way ANOVA and were deemed comparable (p = 0.5508).

Figure 1

Regarding cytotoxicity as assessed using NRU viability in human lung epithelial cells, undiluted cigarette smoke demonstrated full cytotoxicity within 10 undiluted puffs, whereas all THPs elicited a full toxic response only after delivery of 180 undiluted puffs (Figure 1b). On this basis, the THP response was deemed ~ 95% less toxic relative to cigarette smoke (16). To compare the THPs in more detail, we selected a dose point that was in the toxic range of all products (85 puffs) (this dose represented the first dose beyond the half-maximal inhibitory concentration (IC₅₀)). The five THP variants and the THP_base showed no significant difference in toxicity (p = 0.8378) (Figure 1c), suggesting that the top flavours and innovations in the variants had no impact on overall toxicity. As compared with the THP_benchmark, the THP_base and variant THPs were all deemed significantly less toxic relative to cigarette smoke (p = 0.0141).

Lastly, when the cytotoxicity data were presented as a function of total analyte yields (TAY; i.e., the accumulation of all analytes measured), the THP_base and five variants were grouped together, whereas the cigarette data sat independently from those of the THPs (Figure 1d). Interestingly, cigarette smoke reached a cytotoxicity of 80% at a TAY of 20 mg. By contrast, at 50%–60% toxicity, the THPs all delivered a TAY of approximately 300 mg. These data suggest that the cigarette smoke contains chemicals that are more toxic, and a much smaller dose is required to achieve greater cytotoxicity relative to the THP aerosols. This is consistent with the observed cytotoxicity curves (Figure 1b) and the measured emissions, which showed that the THPs are approximately 94% less toxic than a combustible product (21).

The development and the availability of alternative tobacco and nicotine products are fast evolving. New products such as THPs are becoming more acceptable to the consumer and their use is growing globally. As consumer popularity increases, so does innovation and product evolution. This leads to shorter product life cycles, with continued requirement to appropriately assess their safety and quality. As part of a toxicological assessment, products undergo careful risk assessments of their chemical content and emissions, and thresholds of toxicological concern (TTC) are derived. The device components and the potential for degradation and leachable by-products are also assessed (27). Only when the product meets these standards (including battery performance) is the product launched. However, conducting such risk assessments using TTCs is time-consuming, and it is almost impossible to conduct a complete chemical, biological, and clinical review of each new product variant. Thus, as used in the pharmaceutical industry (12), a practical testing strategy is required to demonstrate the ‘equivalence’ of product variants based on an original product with a comprehensive science dataset.

As a first towards demonstrating the feasibility of an equivalence or a bridging approach, in this study we investigated whether five THP variants, relative to the THP_base with a foundational dataset (6, 14,15,16,17,18,19,20,21), maintain a similar percentage reduction in their chemical and in vitro toxicological profile when compared with cigarette smoke. The five THP variants were each classified as having minor (e.g., different top flavours or nicotine strengths) or major (device innovations) changes from the base variant.

Chemical analysis based on the TobReg9 mandated toxicant list (23) and a physiologically relevant aerosol exposure for cytotoxicity assessment, demonstrated that five variant THPs were all comparable to the THP_base, irrespective of innovation to consumable format or top flavour. Relative to cigarette smoke, the reductions in THP aerosol emissions were significantly lower (p < 0.001) and were also comparable among all variants at 94%–97% (Table 1 and Figure 1a). In vitro toxicological analysis of the THP_base aerosol versus cigarette smoke showed an approximately 95% reduction in cell cytotoxicity. The THP_base and five variant THP aerosols showed comparable cytotoxicity using an aerosol-based air-liquid interface approach (p = 0.8378) and were significantly less cytotoxic than cigarette smoke (p = 0.04). When the emissions analysis was combined with the toxicology data, the THP variants showed an approximate 94% reduction relative to the reference combustible product (16). When compared with the THP_benchmark, the THP_base and variant THPs all showed significantly lower toxicity (p = 0.0141).

For a more informative analysis, the emissions data summarized as Total Analyte Yields (TAY) and were plotted against viability (Figure 1d), which showed that the THP_base and variant THPs were grouped together, well apart from the reference cigarette. Thus, cigarette smoke is clearly different from THP aerosol; it is more toxic and requires much lower doses to achieve cytotoxicity. For cigarette smoke, a TAY of only 20 mg was required to achieve 80% toxicity; for THP aerosol, by contrast, a TAY of 300 mg was required to achieve 50%–60% toxicity. Based on TAY calculations from TobReg9 in this study, the chemicals in cigarette smoke are more toxic at lower concentrations as compared with those in THP aerosol. This observation of reduced chemical and toxicological profiles relative to cigarette smoke is consistent with previous reports (7, 15,16,17,18).

A bridging framework will help to support product testing to inform an evidence-based strategy to predict similarity of future product variants to a base product, with an already established body of data. This would therefore avoid repetition of vast data generation and could ease authorisation requirements of newer products. The data presented will help inform potential next steps for a practical bridging framework especially considering the rapidly evolving product landscape, where it will become impossible to assess every variant under a “complete” testing strategy. Nevertheless, the small dataset presented here shows that the five THP variants assessed were all comparable (or “bridgeable back”) to the THP_base product foundational dataset and that the introduction of flavour and/or device innovations did not adversely affect the toxicological impact of the product, under these test conditions using this one approach.

This approach should be caveated in that, only a limited selection of analytes were measured in this proof-of-concept study and therefore the reductions and calculations are only based on a small subset of the total emissions and does not represent the total chemical profile. In addition, the chemicals driving the response may not be captured in this analysis, being based on an accumulation of analytes, the value placed on mg substances may be more heavily weighted compared to those in the ng range. More work is required to understand chemical, biological (in vitro), human consumption, and clinical data before the equivalence of these products can be definitively demonstrated. Future studies may need to assess additional chemical and biological outputs as well as test articles to those presented here, and all data will need to be contextualised against human consumption data in terms of a bridging framework. Figure 2 demonstrates the proposed process in which an expanded bridging framework could follow. This process is evolving and may change as bridging concepts for THPs become more defined. Specifically, data should leverage where possible, genetic toxicological assessments (conducted as part of a stewardship approach in this example), a more comprehensive emissions panel (which included untargeted analysis), combined with in vitro disease and mechanistic screening approaches and human in use data. Finally, to make sense of the growing complexity, bioinformatical approaches should be employed and appropriate competitors such as cigarette smoke and a commercial reference product. We have left this part absent in the framework as this could be project and study specific. Finally, although this study is specific to tobacco heating products, we believe in working towards bridging strategies for new generation products including THPs and electronic nicotine delivery systems.

Figure 2

In conclusion, this study demonstrates the feasibility of a preclinical equivalence concept for tobacco heating products across five variants with shifting base flavour (tobacco, fresh, smooth, and citrus) and combinations of changes (such as foil wrap consumable size and flavour change). The outcomes articulated here are predicated on a sound and robust stewardship approach to ensure product and consumer safety are maintained, in line with published approaches (27). The data presented here is only a small data package to demonstrate the concept of bridging between products and represents the start of a preclinical weight of evidence journey. Work is required to establish a full bridging framework that incorporates additional chemical and in vitro biological data, coupled with human use, and ultimately linked to clinical outcomes. Furthermore, THP responses should be contextualised against cigarette smoke and benchmarked against a commercial comparator (or multiple comparators to give a wider commercial perspective). For the latter we have benchmarked against one that has already been authorised by regulators in the USA.