Demographics and flavor preferences of EC users in the online survey
We conducted an online survey to identify the most popular flavors of EC refill fluids. Of 2,753 participants, 853 were current EC users (Supplemental Table 1). Most EC users were between ages 18–22 (49.5%), male (72.0%), and listed “some college” as their highest education (39.0%). The most represented ethnic groups were White/Caucasian (43.0%), followed by Asian (23.4%), and Hispanic/Latino (19.0%). 87.0% were ever cigarette users. 68.0% were ever cigarette users that no longer smoked. 53.1% listed ECs as aids to quit smoking. 49.0% listed nicotine replacement products as aids to quit smoking. EC use was “influenced by friends” (54.0%) and some believed “vaping is safer than smoking” (55.0%) (Supplemental Table 1). Most users described their health as “very good/good” (73.0%) and vaped daily (82.0%) for at least 1 month to 2 years (74.0%). Most participants described their use of EC as “regularly, at least once a day” (57.0%) and for 31–59 minutes a day (22.0%) or 1–3 hours a day (22.0%). In users’ decisions to vape, refill fluid flavors were deemed “very important” or “important” (78.0%). The most popular models of EC were tanks/mods (47.0%). Over half of the EC users (54.0%) currently used EC products, while 19.0% were former users, and 27.0% were dual users of both EC and conventional cigarettes.
EC users (N = 789) indicated their flavor preferences from 18 possible flavor categories (Fig. 1A). They were able to select more than one flavor. The top six flavor preferences were “Berries/Fruits/Citrus” (N = 559), “Sweet” (N = 406), “Bakery/Dessert” (N = 321), “Mint/Menthol” (N = 298), “Candy” (N = 293), and “Buttery/Cream/Caramel/Vanilla” (N = 274) (Fig. 1A). The two least popular flavors were “Nuts” (N = 44) and “Savory/Dinner Food” (N = 21). Flavor preferences were similar irrespective of the users’ age (Fig. 1B).
Frequency distribution of popular flavors from survey, local shops, and online stores. (A) Results from the online survey. (B) Popularity of flavor categories among different age groups in the online survey. (C) Results from the local and online stores. Frequency on the y-axis refers to the number of times each flavor category (x-axis) appeared in the population.
Popular flavors in local and online shops
To confirm the results of the online survey, 17 EC vape shops in southern California were contacted by telephone or visited to obtain information on their top-selling refill fluids. Each local shop reported 5–10 top-selling refill fluid categories, the majority of which were “Berries/Fruits/Citrus” (54 local; 57 online) (Fig. 1C). The Internet was used to determine flavor profiles when shops could not provide these data. In addition, nine popular online shops were visited, and the flavor profiles for 5–10 top-selling fluids were identified as “Berries/Fruits/Citrus” (Fig. 1C). Flavor categories that were not among the most popular are not included in Fig. 1C.
Based on the above data, 20 top-selling refill fluids were purchased from four shops in Riverside County, CA. One local shop specialized in “cloning” brand-name EC fluids, while the other shops sold products that were made by refill fluid manufacturers. We distinguish these products as “cloned” and “authentic”, respectively. It is important to consider cloned products because they are often less expensive than their authentic branded counterparts, and some shops sell mainly cloned products. Supplemental Table 2 shows the flavor profile and general flavor category for each product purchased in local shops.
Identification and quantification of flavor chemicals in the 20 popular refill fluids
A total of 99 flavor chemicals were identified and quantified in the 20 EC refill fluids purchased in local shops (Fig. 2; Supplemental Table 3). The general flavors associated with each chemical are given in Supplemental Table 3, and the target flavors not found in any of the products are given in Supplemental Table 4. The total concentration of the flavor chemicals in each product, which is given at the top of the columns in the heat map (Fig. 2), ranged from 0.63 mg/ml (“Bird Brains”) to 27.9 mg/ml (authentic “Dewberry Cream”). The x-axis of the heat map is sorted based on the total flavor chemical concentration (highest on the left).
Heat map showing flavor chemical concentrations in 20 popular refill fluids. Chemicals are ordered on the y-axis according to their toxicity (based on LD50 data from oral exposure in rats) and within each class, they are ranked from most to least toxic. Products (x-axis) are ordered according to the weight (mg/ml) of all the flavor chemicals in each product with the highest concentration at the left. Numbers 1–4 with product names denote stores where refill fluids were purchased, and “C” indicates a cloned product. “Rainbow Sherbet” is a clone of “Unicorn Puke” and “Melon Mania” is a clone of “Mega Melons”. The total chemical concentration (mg/ml) and the number of individual chemicals is indicated at the top of each column. Nicotine, which is not a flavoring, is in the bottom row for comparison.
On the y-axis of the heat map, the 99 chemicals were ranked by their safety classification (Toxic, Harmful, Irritant, and No data) as posted on the Good Scents Flavor Company website17, which provides peer-reviewed information for the flavor, food, and fragrance industry. Within each safety classification, chemicals are listed from most to least toxic based on rat oral LD50, also posted on the Good Scents website. For most flavor chemicals, one LD50 value was available, but if multiple were given, we chose the LD50 value reported in the journal of Food and Cosmetics Toxicology. Rat oral data were used for ranking because they were available for most chemicals in the heat map, while inhalation LD50 data were seldom available. The y-axis ranking was useful for predicting which chemicals would be most toxic and therefore most interesting to pursue; however, it is not intended to imply that the chemicals in refill fluids produce the same effects as in the rat oral data. The chemicals with the highest concentrations and highest predicted toxicities are in the upper left quadrant of the heat map.
“Bird Brains” had the fewest flavor chemicals (N = 22), while authentic “Dewberry Cream” had the most (N = 47). In some cases, these chemicals were very low in concentration (e.g., maltol in “Bird Brains”), while in others the concentrations exceeded 1 mg/ml (e.g., ethyl maltol in “Dewberry Cream”). Thirteen percent of the flavor chemicals were present at concentrations higher than nicotine in some samples.
The frequency with which individual chemicals were found in the 20 popular products varied. Some were found in all or almost all refill fluids (e.g., maltol and ethyl acetate), while others were only in 2–3 products (e.g. ethyl lactate and citral) (Fig. 3). Of the 99 chemicals identified in the popular products, 28 appeared in at least 10 of 20 products, indicating that a subset of flavor chemicals is used frequently. Those chemicals that appeared in only one product are shown in Supplemental Table 5.
Frequency distribution of flavor chemicals within popular products and their chemical class. Chemicals are ranked according to their frequency in popular products for all data. The inset shows the class to which each chemical belongs.
Data were also analyzed according to their chemical class (Fig. 3 insert). Most flavor chemicals were esters, and many were terpenes, phenols, alcohols, ketones, and aldehydes. The “other” category included benzopyrone, pyrazine, pyrone, and thiazole. While not shown in Fig. 3, some chemicals belong to more than one class, such as vanillin, which is both an aldehyde and phenol.
In Fig. 4, the flavor chemical data were filtered to include only those refill fluids (17 of 20) that had at least one chemical at a concentration ≥1 mg/ml. Filtering at this level reduced the number of flavor chemicals from 99 to 18, which we further considered in this study.
Heat map of popular EC refill fluids with at least one flavor chemical ≥1 mg/ml. These flavor chemicals were considered to be dominant in the popular refill fluids that were analyzed. They are ranked on the y-axis according to rat oral toxicity and on the x-axis according to total concentration (mg/ml) of the flavor chemicals.
Identification of cytotoxic EC refill fluids
The cytotoxicities of the 20 popular refill fluids were evaluated using the MTT assay, which measures mitochondrial reductase activity. Decreases in the MTT assay relative to untreated controls are indicative of cytotoxicity due to decreases in mitochondrial metabolism and/or cell survival. The concentrations required for a 30% (IC70) and 50% (IC50) reduction in the MTT assay were determined for each refill fluid (Fig. 5; Supplemental Fig. 1). Some products (e.g., “Bird Brains”) showed no cytotoxicity (Fig. 5A; Supplemental Fig. 1A). Most refill fluids (e.g., “Ho!Ho! Watermelon”) reached at least an IC70 (30% inhibition vs control), indicating they were cytotoxic by ISO standard 10993-518 (Fig. 5B). Four refill fluids (“Dewberry Cream”, “Dewberry Cream” clone, “Mega Melons”, and “Kiberry Yogurt”) (Fig. 5C–E) reached at least IC50 values (50% inhibition vs. control), again indicating cytotoxicity. Figure 5E summarizes the cytotoxicity data relative to the untreated control for cells treated with a 1% concentration of each refill fluid. Table 1 shows the IC70 and IC50 values for all 20 products. When tested independently, propylene glycol, glycerol, and nicotine were not cytotoxic at concentrations found in the 1% refill fluid solutions (Supplemental Fig. 2).
Cytotoxic refill fluids identified using mNSC. (A–D) Representative MTT concentration- response curves for products that were: (A) not cytotoxic, (B) cytotoxic reaching IC70, and (C,D), highly cytotoxic reaching IC50. Summary of cytotoxicity screening results showing products that had little effect (green dots), reached an IC70 (blue squares), or reached an IC50 (red triangles). The most cytotoxic products were “Dewberry Cream”, “Dewberry Cream” clone, “Mega Melons”, and “Kiberry Yogurt”. “Rainbow Sherbet” is a clone of “Unicorn Puke” and “Melon Mania” is a clone of “Mega Melons”. Each graph is the mean ± the standard error of the mean for three independent experiments. *p < 0.05, ***p < 0.001, ****p < 0.0001.
Relationship between cytotoxicity and the total number and total concentration of flavor chemicals
Cytoxicity was examined as a function of the total number of flavor chemicals (Fig. 6A) and total concentration of flavor chemicals (Fig. 6B) in each product. The correlations (R2) between cytotoxicity and the total number of flavor chemicals in a refill fluid or the total concentration of flavor chemicals in each product were 0.42 and 0.54, respectively. The p values of the correlation coefficients were 0.002 (Fig. 6A) and 0.0002 (Fig. 6B), indicating they were statistically significant.
The relationship between cytotoxicity and the total number of flavor chemicals and the total concentration of flavor chemicals. Cytotoxicity is plotted as a function of the total number of flavor chemicals (A) and the total concentration of flavor chemicals (B) in each of the popular refill fluids. Green dots indicate refill fluids that were not significantly cytotoxic, blue dots are refill fluids that reach an IC70, and red dots are refill fluids that reached an IC50. Letters with each point correspond to the products listed in Figs 8 and 9.
Identification of cytotoxic flavor chemicals
Figure 7A allows a direct comparison of the two most cytotoxic authentic refill fluids and their corresponding clones. Total flavor concentrations in “Dewberry Cream” (27.9 mg/ml) and its clone (20.17 mg/ml) were similar; however, “Mega Melons” (authentic) had a higher total flavor concentration (14.59 mg/ml) than its clone (“Melon Mania”) (3.96 mg/ml). In no case were the flavor chemicals in the clones an exact match in number or concentration to their authentic counterpart.
Chemicals in “Dewberry Cream” and “Mega Melons” and their cytotoxicity. (A) Heat map showing the flavor chemicals and their concentrations in the two most cytotoxic refill fluids and their clones. Chemicals are ordered on the y-axis according to their toxicity and within each class, they are ranked from most to least toxic. Products (x-axis) are ranked according to the total flavor chemical concentration, with the highest on the left. The total flavor chemical concentration and number of individual flavor chemicals are indicated at the top of the heat map. Nicotine is in the bottom row for comparison. (B–G) Concentration-response curves of authentic standard chemicals present in the highest concentrations in the two most toxic refill fluids and their clones. (H–K) Concentration-response curves for four chemicals frequently used or present at over 1 mg/ml in refill fluids. Each graph is the mean ± the standard error of the mean for three independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. HD = high dose tested.
We hypothesized that chemicals that were high in concentration in the upper region of Fig. 7A would contribute to the cytoxicity observed in the MTT assay. Six chemicals (ethyl maltol, maltol, vanillin, ethyl vanillin, benzyl alcohol, and furaneol) were >1 mg/ml in “Dewberry Cream” and/or “Mega Melon” and slightly lower in the less toxic clones. Authentic standards of these chemicals were tested in the MTT assay using mNSC and human BEAS-2B cells (Fig. 7B–G). The highest concentration for each authentic standard was chosen to match the highest concentration found in authentic “Dewberry Cream” and “Mega Melons”. In support of our hypothesis, all six authentic standards were cytotoxic at the concentrations found in the refill fluids. mNSC were slightly more sensitive than BEAS-2B to ethyl maltol and benzyl alcohol. For all other chemicals, concentration-response curves were similar for the two cell types. Based on the IC50 data, the chemicals that were the most toxic from high to low were: ethyl maltol, furaneol, maltol, ethyl vanillin, vanillin, and benzyl alcohol.
Although their predicted toxicities based on the rat-oral data were lower than the chemicals in the above assays, ethyl butanoate, triacetin, acetoin, and ethyl acetate, were evaluated in a secondary MTT screen (Fig. 7H–K). Both ethyl butanoate and triacetin were cytotoxic at the highest concentrations found in the 20 products, while the other two chemicals were not cytotoxic.
Figure 8 shows the cytotoxicity for each of the refill fluids at 1%, the concentration of each flavor chemical at 1%, and the cytotoxicity of each flavor chemical based on the authentic standard data. In general, when the parent refill fluid was cytotoxic at 1%, it contained flavor chemicals that could account for its cytotoxicity (e.g., “Dewberry Cream” had a cytotoxic level of ethyl maltol). Exceptions to this, such as “North Shore”, which was cytotoxic (51% of control) but did not have a cytotoxic level of flavor chemicals, suggest that our target list of chemicals does not contain some of the flavor chemicals that are used in refill fluids, that chemicals act additively or synergistically to produce cytotoxicity, or that another factor, such as a metals19, caused cytotoxicity in this product.
Concentrations (mg/ml) of flavor chemicals in 1% refill fluids and their cytotoxicity. Color code indicates the cytotoxicity of flavor chemicals at the concentrations found in 1% refill fluids. Magenta = concentrations that would reach an IC50; Light pink = concentrations that would reach an IC70, Blue = no cytotoxic effect. ND = indicates chemical was not detected in the GC-MS analysis. Code alphabet colors match summary of cytotoxicity of popular refill fluids in the MTT assay (Fig. 5E). Red = 0–50%, Blue = 51–69%, and Green = 70–100%. Flavor names: EM = ethyl maltol; F = furaneol; M = maltol; EV = ethyl vanillin; V = vanillin; BA = benzyl alcohol; EB = ethyl butanoate; T = triacetin; A = acetoin; EA = ethyl acetate.
Since refill fluids are used in ECs without dilution (100%), Fig. 9 is included to show the actual concentration of each flavor chemical in the undiluted parent refill fluid and the cytotoxicity that would be predicted for each chemical at the actual concentration used by EC vapers to produce aerosol. At actual flavor concentrations, all refill fluids would be predicted to be cytotoxic. Ethyl maltol, furaneol, and maltol were always present at concentrations that would be cytotoxic, and these three chemicals were used frequently (maltol for example was in 18 of 20 products tested).
Projected cytotoxicity of flavor chemicals at concentrations (mg/ml) found in refill fluids. Color code indicates the projected cytotoxicity of flavor chemicals at the concentrations found in refill fluids. Magenta = concentrations that would reach an IC50; Light pink = concentrations that would reach an IC70, Blue = no cytotoxic effect. ND = indicates chemical was not detected in the GC-MS analysis. Flavor names: EM = ethyl maltol; F = furaneol; M = maltol; EV = ethyl vanillin; V = vanillin; BA = benzyl alcohol; EB = ethyl butanoate; T = triacetin; A = acetoin; EA = ethyl acetate.
Relationship between the cytotoxicity of each refill fluid at a 1% concentration and each authentic standard chemical
Each highly cytotoxic refill fluid contained one or more of the six toxic flavor chemicals at concentrations that were as cytotoxic as authentic standards in the MTT assay (Fig. 8). In general, moderately cytotoxic refill fluids had lower concentrations (e.g., “Lava Flow” and “WTF” clone) or non-cytotoxic concentrations (e.g., “Blueberry Hills” and “Unicorn Puke”) of the six toxic chemicals. Non-cytotoxic products or those that did not differ from the control by more than 30% either had none of the six toxic flavor chemicals (e.g., “Bird Brains”) or had low levels (e.g., “Overnight”). These data demonstrate a positive relationship between the concentration of ethyl maltol and the cytotoxicity of the refill fluids in which it was used.
The cytotoxicity of refill fluids (1% concentration) was plotted as a function of the flavor chemical concentration in each fluid at 1% (Fig. 10A–J). Dots are color-coded to toxicity of the refill fluids (red = highly cytotoxic, blue = moderately cytotoxic, green = non-cytotoxic), and the letter code with each dot correlates to a refill fluid in Fig. 8. The cytotoxicity and ethyl maltol concentrations in each fluid were highly correlated (R2 = 0.93; p value = < 0.0001). This high correlation occurs because ethyl maltol was the most cytotoxic of the chemicals tested and it maintained its toxicity when tested in a refill fluid. The correlation coefficient was also significant for ethyl vanillin (R2 = 0.68; p value = 0.0033), maltol (R2 = 0.502; p value = 0.0010) and vanillin (R2 = 0.49; p value = 0.0028), but decreased and was not significant for the remainder of the toxic chemicals (Fig. 10E–J).
Relationship between the cytotoxicity of each refill fluid at 1% concentration and the concentrations of each authentic standard chemical. Green dots indicate refill fluids that were not significantly cytotoxic, blue dots are refill fluids that reach an IC70, and red dots are refill fluids that reached an IC50. Letters associated with dots correspond to products in Fig. 8. Because refill fluids are mixtures of cytotoxic chemicals, only ethyl maltol (the most toxic of the authentic standards) had a high correlation coefficient. The p values for ethyl maltol, maltol, ethyl vanillin and vanillin indicate that the correlations are statistically significant. Correlation coefficients for the other chemicals were affected by the presence of ethyl maltol.
