Flavor chemical concentrations in refill fluids often exceeded concentrations permitted in other consumer products. These data support the regulation of flavor chemicals in EC products to reduce their potential for producing both cancer and non-cancer toxicological effects.Adverse health effects have been linked to electronic cigarette use in prior experimental studies on cells, animals, and humans, case reports,and Internet posts. The recent epidemic of “electronic cigarette or vaping product use associated lung injury” has further heightened concerns about the safety of ECs. The Centers for Disease Control and Prevention suggested that poor quality counterfeit and black-market products are linked to some EVALI cases10 and further recommended that vaping products not be used until the causes of EVALI are determined. 11 We have previously shown that some EC refill fluids contain very high concentrations of some flavor chemicals and that the presence of some flavor chemicals at high levels is significantly correlated with cytotoxicity. Although flavor chemicals have not been directly linked to EVALI, we did previously conclude that the high concentrations of flavor chemicals used in some EC refill fluids may cause adverse health effects. While many flavor chemicals in EC products are GRAS for ingestion; their safety has not been evaluated for inhalation. 16 Some EC products have flavor chemical concentrations that far exceed those acceptable for ingestion, for example, we have found cinnamaldehyde in one product at 343 mg/mL. Most prior studies on EC flavor chemicals have been done using products purchased in one country, often the USA,square pot and have generally focused on identification only. In this study, all products were manufactured by one company, and purchases were made in four different countries.
We compared the flavor chemicals in each product to determine: if there were variations in content and concentration with country, if products were cytotoxic, if specific flavor chemicals contributed to cytotoxicity, if any flavor chemicals or co-constituents were present in high enough concentrations to be a risk factor for cancer and how flavor chemicals in the current study compared to those we have examined previously. Mouse neural stem cells are sensitive to EC refill fluids, are amenable to highthroughput screening, and are an excellent model for neurological development. mNSC werecultured in Nunc T-25 tissue culture flasks containing growth medium prepared using methods previously described.For the MTT experiments, cell concentrations were determined using a BioMate 3S Spectrophotometer -based standard curve, and single cells were plated at 1500 cells/well in 96-well plates. For live-cell imaging in a BioStation CT , mNSC were seeded at 5000 cells/well in 24-well uncoated culture plates and allowed to attach overnight before imaging. Seeding densities were adjusted to achieve ~80-85% confluency at the end of the experiments. Human bronchial epithelial cells , which are often used in inhalation toxicology studies, were cultured in bronchial epithelial cell growth medium using protocols previously described.At 80% confluency, cells were harvested and plated at 3500 cells/well in pre-coated 96-well plates for the MTT assay.Direct effects of EC refill fluids or authentic standards of flavor chemicals on mitochondrial reductases were evaluated in concentration-response experiments that included untreated wells to control for vapor effects. After seeding and overnight attachment, cells were either treated with 0%, 0.001%, 0.1%, 0.03%, 0.1%, 0.3%, and 1% refill fluids solutions or 10 fold dilutions of the actual concentration of authentic standard solution made up in culture medium.
All treatments were incubated for 48 hours at 37 ºC. After treatment, 3–2,5- diphenyltetrazolium bromide reagent was added to wells and incubated for 2 hours at 37ºC. Solutions were removed from wells, and 100 µl of dimethyl sulfoxide were added to each well to solubilize formazan crystals. Absorbance readings were taken against a DMSO blank at 570 nm using an Epoch microplate reader . For each variable tested, three independent experiments were performed.For non-invasive analysis of cell morphology, motility, and survival, live-cell imaging was performed using a 10x phase contrast objective in a BioStation CT using automatic Z-focus. After attachment, mNSC were treated with refill fluid solutions at 0.1%, 0.3% and 1% made up in culture medium. Images were taken at 5 – 8 regions in each well once every 2 hours for 48 hours to collect time-lapse data for analysis. Evaluation of mNSC confluency, morphology, and survival was compared in control and treated groups using CL Quant software . For GC/MS analysis, each sample was analyzed twice, and the means were plotted using Prism software . For the MTT assay, data were normalized to the negative control , and treatment groups were expressed as percentages of the negative control. IC50s were computed using the log inhibitor vs. normalized response-variable slope in GraphPad Prism, and IC70s were evaluated visually. Statistical significance in the MTT assay was determined using a one-way analysis of variance , and when there was significance, treated groups were compared to the untreated control. In the live-cell imaging assay, significance was evaluated using a two-way ANOVA in which the variables were time and treatment. The number and concentrations of flavor chemicals in 105 refill fluids were evaluated . Each refill fluid was grouped into a product flavor category and compared for variability based on country of purchase. Refill fluid categorization was done according to flavors and types on the manufacturer’s website . Products are sorted from lef to right in Figure 1 in order of decreasing total concentrations of flavor chemicals.
Based on our target analyte list, the total number and concentration of flavor chemicals varied among products. Two “Q American Blend Tobacco” products did not have any chemicals on our target analyte list. Total flavor chemical concentration and number in original LIQUA flavors were high in “Two Apple” and “Peach” , “Mints,” and Two Mints” , and “RY4 Tobacco” , and “Sweet Accelerator” , and Cheesecake . Within the mint/menthol groups, the total concentration of flavor chemicals varied with “Mints” , having over twice the total concentration of the other products . In all these products, total flavor chemical concentration was > 10 mg/mL, and the total number of flavor chemicals was > 10. In contrast, low total concentrations of flavor chemicals were found in various categories . Based on the duplicate samples we processed, the total number of flavor chemicals and their concentrations were similar in most products with the same flavor name irrespective of country of origin . However, there were some exceptions, such as “Apple” , which was purchased in different cities within the USA and had different flavor chemical concentrations. The concentrations of flavor chemicals across all products ranged from 0.001 – 44.3 mg/mL . All products with ≥ 10 mg/mL in total flavor chemicals contained 3-9 dominant flavor chemicals , and the most frequently occurring were ethyl maltol, triacetin, corylone, ethyl vanillin, vanillin, and menthol . When comparing flavor chemical concentrations across duplicate products purchased in different countries, concentrations of specific chemicals were generally similar . However, we did find some differences. For example, the concentration of corylone was about five times lower in the “Peach” product purchased in the UK than in those from the two US sites and China. Moreover, for “Ry4 Tobacco”, the concentrations of corylone and furaneol varied with the location of purchase. The frequency of occurrence of the 126 flavor chemicals is shown in Figure 3a and Supplemental Figure 2. In descending order of frequency,blueberries in containers the most frequently used flavor chemicals, which appeared in at least 30 products, were triacetin , ethyl butanoate , ethyl maltol , γ-decalactone and δ-decalactone , hydroxyacetone , vanillin and ethyl acetate , 3-Hexen-1-ol and linalool , corylone , and phenethyl alcohol . Less frequently used flavor chemicals that appeared in fewer than 6 products are shown in Supplementary Figure 2. Using publicly available safety information, 22 flavor chemicals were grouped according to their potential to cause harm . Most of the flavor chemicals identified were either “irritants” or “harmful” . At the same time, two were “irritant and dangerous to the environment” , 2 were “harmful and dangerous to the environment” , and one was “toxic” . Additional information on flavor chemicals less frequently used is included in Supplementary Figure 2. Esters, terpenes, and ketones were the most abundant chemical classes .The cytotoxicity of 16 refill fluids that contained at least one flavor chemical ≥ 1 mg/mL and total flavor chemical concentrations ≥ 10 mg/mL is shown in Figure 4. The MTT assay, which evaluates the metabolic activity of mitochondria, was performed using mNSC and BEAS-2B cells after 48 hours of exposure to dilutions of refill fluids in submerged culture. Absorbances that are lower than the untreated controls indicate that the treatment decreased mitochondrial reductase activity. Cytotoxic refill fluids and their inhibitory concentrations at 70 % and 50 % are shown in Figure 4 and Table 2.
“Two Apples” and “Ry4” were the most cytotoxic refill fluids and duplicates from multiple countries produced similar results in the MTT assay. When cytotoxicity was observed, the mNSCs were generally more sensitive to the effects of the refill fluids than the BEAS-2B cells. Even though “Cheesecake,” “Peach,” “Mints” and “Honeydew” contained relatively high concentrations of flavor chemicals, they produced little to no response in either cell type in the MTT assay. Non-invasive analysis of mNSC growth was performed using time-lapse images of cells taken over 48 hours. “Two Apples” from Nigeria and China, and “Ry4” from the USA and China inhibited cell growth in a concentration-dependent manner irrespective of country of origin . In the treatment group with “Two Apples,” 2-way ANOVA revealed statistical significance as early as 12 hours and 20 hours for cells treated with EC refill fluid solutions at 1% and 0.3% . The effect observed when cells were treated with 0.1% solutions was statistically different from the control starting at 34 hours . Micrographs show images taken at 0, 24, and 48 hours. Compared to the untreated group, 0.3 and 1 % concentrations inhibited cell growth early in the experiment . The effects of “Ry4 Tobacco” on mNSC growth at 1% and 0.3% were similar with p values < 0.0001 starting at 10 hours . 0.1% differed significantly from the control beginning at 20 hours . Peach did not significantly alter growth in any treatment . To evaluate the effects of authentic standards of flavor chemicals individually and as mixtures, BEAS-2B cells were treated with concentrations of specific flavor chemicals that were dominant in “Peach” and “Mint” . Inhibitory concentrations at 70 % and 50 % which are indicators of cytotoxicity23 are shown in Table 3. Individually, triacetin , corylone , and γ-decalactone at the concentrations found in “Peach” would be cytotoxic to BEAS-2B cells. However, when combined, there was no effect in the MTT assay . Similarly, the concentrations of triacetin and carvone in “Mint” are high enough to induce significant cytotoxic effects individually, but when combined, the mixture was non-cytotoxic .Regression analysis was performed to determine if cytotoxicity correlated with total flavor chemical concentrations , the total number of flavor chemicals , and the concentration of individual flavor chemicals . The correlations were grouped into 3 categories: high , moderate , and low . Cytotoxicity was strongly correlated with total flavor chemical concentration for mNSC and moderately correlated for BEAS-2B cells . The relationship between the total number of flavor chemicals and cytotoxicity was moderate for BEAS-2B and not correlated for mNSC . The concentrations of six flavor chemicals were high to moderately correlated with cytotoxicity for both cell types . Although carvone was not very cytotoxic in the MTT assay, its concentration did correlate with cytotoxicity for mNSC cells, but not for BEAS-2B cells . Triacetin concentrations, which were high in “Peach” flavored products, were not correlated with cytotoxicity for BEAS-2B cells or mNSC .Some refill fluid chemicals are known or probable carcinogens. The Margin of Exposure approach aids risk managers in prioritization and is used by the FDA and other expert groups to access the cancer risk of food additives. 24-27 The MOE is the ratio of a reference point for an adverse effect to the estimated daily intake or exposure of a chemical in humans. Reference points obtained from experimental or epidemiological data based on dose-response curves include the BenchMark Dose , the No Observed Adverse Effect Level , or the Low Observed Adverse Effect Level .