Article Text

Download PDFPDF

Comparing the cytotoxicity of electronic cigarette fluids, aerosols and solvents
  1. Rachel Z Behar1,2,
  2. Yuhuan Wang2,
  3. Prue Talbot2
  1. 1 Cell Molecular and Developmental Biology Graduate Program, University of California, Riverside, California, USA
  2. 2 Department of Cell Biology and Neuroscience, University of California, Riverside, California, USA
  1. Correspondence to Dr Prue Talbot, Department of Cell Biology and Neuroscience, University of California, Riverside 92521, CA, USA; talbot{at}


Background As thousands of electronic cigarette (e-cigarette) refill fluids continue to be formulated and distributed, there is a growing need to understand the cytotoxicity of the flavouring chemicals and solvents used in these products to ensure they are safe. The purpose of this study was to compare the cytotoxicity of e-cigarette refill fluids/solvents and their corresponding aerosols using in vitro cultured cells.

Methods E-cigarette refill fluids and do-it-yourself products were screened in liquid and aerosol form for cytotoxicity using the MTT (3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide) assay. The sensitivity of human pulmonary fibroblasts, lung epithelial cells (A549) and human embryonic stem cells to liquids and aerosols was compared. Aerosols were produced using Johnson Creek’s Vea cartomizer style e-cigarette.

Results A hierarchy of potency was established for the aerosolised products. Our data show that (1) e-cigarette aerosols can produce cytotoxic effects in cultured cells, (2) four patterns of cytotoxicity were found when comparing refill fluids and their corresponding aerosols, (3) fluids accurately predicted aerosol cytotoxicity 74% of the time, (4) stem cells were often more sensitive to aerosols than differentiated cells and (5) 91% of the aerosols made from refill fluids containing only glycerin were cytotoxic, even when produced at a low voltage.

Conclusions Our data show that various flavours/brands of e-cigarette refill fluids and their aerosols are cytotoxic and demonstrate the need for further evaluation of e-cigarette products to better understand their potential health effects.

  • electronic cigarettes
  • flavors
  • cytotoxicity
  • vegetable glycerin
  • refill fluid
  • aerosol

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The variety of electronic cigarette (e-cigarette) refill fluids is increasing rapidly,1 and e-cigarettes have become popular with teens as well as adults who previously did not use tobacco products.2 This rapid rise in e-cigarette popularity has occurred with little information on their safety and health risks. A recent risk assessment concluded that additional work is needed to better understand the public health impact that e-cigarettes pose.3

Cig-alike models of e-cigarette come with prefilled cartomizers that contain a solvent, such as propylene glycol or glycerin (also called vegetable glycerin or glycerol), flavourings and nicotine (often ranging from 0 to 36 mg/mL).4 5 Users also purchase bottles of refill fluid that can be manually dripped into e-cigarette cartomizers. Some refill products, referred to as do-it-yourself (DIY), are sold as concentrates that can be diluted by the user. Refill fluids are available in numerous flavours, are commonly customisable in nicotine concentration and solvents, and are often more cost-effective than prefilled cartomizers.

Given their rise in popularity, it is critical to understand the positive and negative health effects that e-cigarettes introduce. Over 25 case reports attribute adverse health effects to e-cigarettes.6 These include systemic effects involving the respiratory, cardiac and digestive systems, unintentional and intentional poisonings, and injuries due to explosion. Several in vitro studies found that e-cigarettes lead to activation of cell stress pathways including inflammation, oxidative stress, apoptosis and DNA damage.7–12 In addition, many negative and some positive symptoms have been attributed to e-cigarette exposure by users self-reporting on Internet forum websites.13 14 The Food and Drug Administation (FDA) has documented minor to severe adverse events reported to them by e-cigarette users.15 16 The above adverse effects may be attributed to nicotine withdrawal or overdose,6 13 metal particles and nanoparticles in e-cigarette aerosol,17–19 reactive oxygen species, or flavourings such as diacetyl and cinnamaldehyde.20 21 Several studies found carbonyl compounds, such as acrolein and formaldehyde, in e-cigarette aerosol.22–24 These highly reactive and toxic compounds are produced by the pyrolysis of the solvents (eg, propylene glycol or glycerin).22 25 26

Our laboratory recently showed that the cytotoxicity of 33 e-cigarette refill fluids and three DIY products varied significantly when tested with adult and embryonic cells.27 The cytotoxicity of some products was correlated with the number and concentration of the chemicals used to flavour the refill fluids. Cinnamon-flavoured products were particularly cytotoxic, and cinnamaldehyde was identified as the most potent additive in these fluids.21 We also reported that cinnamaldehyde is widely used in refill fluids, including popular fruity and sweet flavours, and that it produces adverse effects on cells at doses that do not cause cell death.28

The purpose of this study was to follow-up on our prior publication dealing with the cytotoxicity of refill fluids.27 Specifically, we compare the cytotoxicity of these refill fluids and their corresponding aerosols using three different cell types and also evaluated the cytotoxicity of aerosols made from authentic propylene glycol and glycerin, the two most commonly used refill fluid solvents.


Sources of e-cigarette products

Thirty-six e-cigarette refill fluids were tested in a previous screen27 with the exception #3 Marcado (Johnson Creek), which was replaced with a duplicate bottle (#73 Marcado, Johnson Creek). The tested products were manufactured by Freedom Smoke USA (Tucson, Arizona, USA), Global Smoke (Los Angeles, California, USA), Johnson Creek (Johnson Creek, Wisconsin) and Red Oak (a subsidiary of Johnson Creek). Thirty five of the original 36 bottles of refill fluids and e-cigarette DIY products containing various flavourings and nicotine concentrations were evaluated (table 1). The products were chosen to give a range of manufacturers, solvents, nicotine concentrations and flavours. All bottles were given an inventory number and stored at 4°C for 1 year before rescreening. Manufacturers labelled their products ‘vegetable glycerin’ and/or ‘glycerol’, which are chemically the same. We refer to both terms as ‘glycerin’ when discussing them in the text, but use the manufacturers’ terms (glycerol and vegetable glycerin) in table 1 to show actual labelling of each product.

Table 1

E-cigarette product information and human pulmonary fibroblast cytotoxicity data of the fluids and aerosols

E-cigarette aerosol collection

E-cigarette aerosol was produced using a smoking machine described previously.28 The puffer box was connected with Cole Parmer MasterFlex Tygon tubing (Vernon Hills, Illinois, USA) to a MasterFlex peristaltic pump (Barnart Company, Barrington, Illinois, USA; Model #7520–00). The line between the smoking machine and the pump contained a T connector (Fisher Scientific) that held the e-cigarette. The peristaltic pump was warmed up for a minimum of 15 min before collecting aerosol into culture medium in a round bottom flask submerged into an ice bath.

Each batch of aerosol was prepared with a fresh unused cartomizer. Empty cartomizers were purchased from Johnson Creek for use with their 2.85 V Vea model e-cigarette. To avoid dry puffing, 1 mL of refill fluid was pipetted into each cartomizer as recommended by the vendor, and weights were taken before and after puffing to monitor fluid consumption. The peristaltic pump speed was reduced to zero until just before every puff at which time pump speed was increased to the desired level. The smoking machine was calibrated to draw a puff volume of 30 mL for a duration of 4.3 s, which is the average puff duration for e-cigarette users29 at a frequency of 10 puffs/hour. For each batch of aerosol, 24 puffs were collected into 4 mL of culture medium.

Cell culture

The human pulmonary fibroblasts (hPFs) were chosen as they were used in our previous studies, they are one of the first cell types exposed to inhaled aerosol and they are involved in development of lung diseases, such as chronic obstructive pulmonary disease.30–32 hPFs (ScienCell, Carlsbad, California, USA) were cultured using the manufacturer’s protocol in complete fibroblast medium containing 2% fetal bovine serum, 1% fibroblast growth serum and 1% penicillin/streptomycin. hPFs were dispersed into single cells and plated at 4000 cells/well in a 96-well plate using a BioMate 3S Spectrophotometer (Thermo Fisher Scientific, Chino, California, USA) based standard curve.

A549 lung epithelial cells (ATCC CCL-185 cells, Manassas, Virginia, USA) are often used in toxicological and inhalation testing. A549 cells were cultured using the distributor’s protocol in ATCC F-12K medium and 10% fetal bovine serum using tissue culture flasks. A549 cells were plated as single cells at a density of 50 000 cells/well in 96-well plates using a BioMate 3S Spectrophotometer-based standard curve.

The pluripotent human embryonic stem cells (hESCs) were used as a model for early postimplantation human embryos.33 The hESCs (H9) (WiCell, Madison, Wisconsin, USA) were cultured on Matrigel in mTeSR medium (Stem Cell Technologies, Vancouver, Canada) in six-well plates as described in detail previously.34 For experiments, colonies of 2–10 cells were plated using a standard curve-based method where 40 000 cells/well in a 96-well plate were measured.

Each cell type was plated at a density that grew to about 80% confluency by the end of the experiment.

MTT assay

E-cigarette fluids and their aerosols were tested in 96-well plates in dose–response experiments using the MTT (3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide) assay to observe cytotoxic effects on hPF, hESC and A549 cells. Dilutions of aerosols were tested at concentrations ranging from 0.0006 to 6 TPE (TPE = total puff equivalents, which is the number of puffs/mL of culture medium). To compare non-aerosolised and aerosolised refill fluid, TPE were converted to percentage (%) using the density of the fluid and the weight/puff of the aerosol. Density was determined by comparing the weight of the fluid with that of water on an analytical balance. Then, the weight of one puff of e-cigarette aerosol was calculated by subtracting the weight of the cartomizer before use and after dividing the weight by 24 (number of puffs taken). The weight of one puff was divided by the density of the fluid and converted to a percentage using the total number of puffs/mL of medium.

Thirty-six refill fluids and DIY e-cigarette products were screened for toxicity in a previous study.27 To determine if these products showed similar toxicity after storage at 4°C, dilutions of the 35 refill fluids and DIY products were rescreened without aerosolisation using the MTT assay at concentrations of 0.001%, 0.01%, 0.03%, 0.1%, 0.3% and 1.0%. To make dilutions, all fluids were brought to room temperature, and mixed thoroughly by pipetting.

The MTT assay was performed as described in detail previously.35 36 Refill fluids and aerosols diluted with culture medium were tested in 96-well plates with negative controls located to the left and to the right of the concentrations tested. The control adjacent to the highest concentrations checked for vapour effects produced by volatile fluids and aerosols.35 37 When a vapour effect was found, a lower high concentration was used to rescreen. Cells and treatments were plated together and incubated for 48 hours, and then examined microscopically to observe the condition of the cells. The MTT solution was added for 2 hours, and the MTT assay was performed. Aerosols were tested in three independent experiments, and means and SE of the mean were used to produce dose–response curves for each cell type. Refill fluids were rescreened a single time to observe their correspondence with the original screen.27 In some cases, an additional high concentration (>1%) of refill fluid was tested to allow comparison of the refill fluid with the highest concentration of aerosol. These additional high concentrations are represented as blue block arrows in figures 1 and 2 and supplementary file S2.

Supplementary File 1

Both the refill fluid and aerosol were non-cytotoxic. Red lines are refill fluids previously screened with hPF.27 Blue lines are the same refill fluids (RF) rescreened in this study to determine if storage effected cytotoxicity. Green lines are aerosols produced from each refill fluid and used to treat hPF. For green lines, data are the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different and represent the LOAEL. *p<0.05, ** = p<0.01, *** = p<0.001.
Figure 1

Four patterns of cytotoxicity for e-cigarette refill fluids (RF) and aerosols. (A and B) Both the RF and aerosol were non-cytotoxic. (C and D) Both the RF and the aerosol were cytotoxic. (E) The RF was cytotoxic but the aerosol was not. (F and G) The aerosol was cytotoxic but the RF was not. Red lines are RF previously screened with human pulmonary fibroblast (hPF).27 Blue lines are the same RF rescreened in this study to determine if storage effected cytotoxicity. Green lines are aerosols produced from each RF and used to treat hPF. For green lines, data are the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different from the untreated controls (lowest observed adverse effect level). *p<0.05, **p<0.01, ***p<0.001.

Figure 2

Most e-cigarette products containing glycerin produce cytotoxic aerosols. (A–L) Comparison of the 11 refill fluids (RF) and one do-it-yourself purchased sample of vegetable glycerin. Red lines are RF previously screened with human pulmonary fibroblast (hPF).27 Blue lines are the same RF rescreened in this study to determine if storage effected cytotoxicity. Green lines are aerosols produced from each RF and used to treat hPF. For green lines, data are the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different and represent the lowest observed adverse effect level. *p<0.05, **p<0.01, ***p<0.001.

Data analysis

For aerosol dose–response experiments, inhibitory concentration at 50% (IC50) values were computed with Prism software (GraphPad, San Diego, California, USA) using the log inhibitor versus normalised response-variable slope with the top and bottom constraints set to 100% and 0%, respectively. For the refill fluid redo screen, IC50 values were determined by eye from the dose–response curve. Statistical significance for aerosols was determined using an analysis of variance on three independent experiments using GraphPad Prism. When significance was found, treated groups were compared with the lowest concentration using Dunnett’s post hoc test, and means were considered significantly different for p<0.05. The no observed adverse effect levels (NOAEL) and the lowest observed adverse effect levels (LOAEL) were determined by statistical significance.


Cytotoxicity of refill fluid and aerosol samples with hPF

Table 1 summarises data on the products used, our inventory numbers, the company of origin, the amount of nicotine/product, the solvent when known, the flavour category, the dose–response data (NOAEL, LOAEL and IC50) and the cytotoxicity data (response <70% of the control) for both the fluids and their corresponding aerosols. All of the products were considered refill fluids except for #24, #32 and #33, which were DIY products. These products ranged in potency (table 1), and all dose–response curves for the 35 aerosols, with one exception, yielded concentrations that were significantly more toxic than the lowest concentration tested. While flavouring categories, such as creamy/buttery, mint/menthol, tobacco and fruit were within the hierarchy of potency, 6 of the 14 ‘creamy’ aerosols (#4, #19, #26, #28, #29, #30) were the most potent in this screen.

Generally, unheated refill fluids produced similar dose–response data to that obtained in our original screen (supplementary files S1-3), showing that cytotoxicity of the refill fluids did not change measurably when stored for prolonged times at 4°C. Clear exceptions to this are #34 JC Original by Johnson Creek (supplementary file S1G) and #40 Caramel by Global Smoke (supplementary file S2R), which lost potency with storage. Differences between the previous screen and current rescreen may be due to ageing of the product or escape of volatile flavouring chemicals during storage.

Supplementary File 2

Both the refill fluid and the aerosol were cytotoxic. Red lines are refill fluids previously screened with hPF.27 Blue lines are the same refill fluids (RF) rescreened in this study to determine if storage effected cytotoxicity. Green lines are aerosols produced from each refill fluid and used to treat hPF. For green lines, data are the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different and represent the LOAEL. *p<0.05, ** = p<0.01, *** = p<0.001.

Sensitivity of adult respiratory cells and stem cells

A subset of aerosols from the refill fluids that were most cytotoxic in the original screen was further tested using hESC and A549 human lung epithelial cells (figure 3). hESCs were more sensitive than hPF and A549 cells to about 50% and 40% of the aerosols, respectively. This agrees with our previous data showing that stem cells are more sensitive than differentiated adult cells to e-cigarette products.27 The two adult lung cell types responded similarly to 80% of the aerosol treatments.

Figure 3

Two adult respiratory cell types and a pluripotent cell compared for sensitivity to e-cigarette aerosols. H9 hESC (red lines), A549 cells (blue lines) and hPF (green lines) were treated with 10 e-cigarette aerosols (A–J). Data are plotted as the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different from the untreated control (lowest observed adverse effect level). *p<0.05, **p<0.01, ***p<0.001. hESC, human embryonic stem cell; hPF, human pulmonary fibroblast.

Patterns of cytotoxicity

The dose–response data from the above screen can be broken down into four patterns of cytotoxicity by determining whether the highest concentration was less or greater than 70% of the control (figure 1; supplementary files S1-3).38 For example, if the sample’s high concentration was <70% of the control, the refill fluid or aerosol was considered cytotoxic. The patterns were characterised as follows: (1) both the refill fluid and its aerosol were non-cytotoxic (7 of 35=20%) (figure 1A,B and supplementary file S1); (2) both the refill fluid and its aerosol were cytotoxic (19 of 35=54%) (figure 1C, D and supplementary file S2); (3) the refill fluid was cytotoxic but the aerosol was not (1 of 35=3%) (figure 1E); and (4) the aerosol was cytotoxic but the refill fluid was not (8 of 35=23%) (figure 1F, G and supplementary file S3). Collectively, when combining patterns (1) and (2), 74% of the 35 unheated refill fluids/DIY products correctly predicted the cytotoxicity of their corresponding aerosol. In some cases, either the refill fluid or aerosol was just above 70% of the control, while the other fell just below 70%. For these graphs, the percent difference was calculated and if the two numbers were within 10%, they were considered both non-cytotoxic.

Supplementary File 3

The aerosol was cytotoxic but the refill fluid was not. Red lines are refill fluids previously screened with hPF.27 Blue lines are the same refill fluids (RF) rescreened in this study to determine if storage effected cytotoxicity. Green lines are aerosols produced from each refill fluid and used to treat hPF. For green lines, data are the means and SE of the mean of three independent experiments. Asterisks indicate the lowest concentrations that are significantly different and represent the LOAEL. ** = p<0.01, *** = p<0.001.

Most refill fluids/DIY products containing only glycerin produced cytotoxic aerosols

Ten of the 11 aerosolised glycerin-based refill fluids (91%) were cytotoxic (figure 2A-K). Aerosolised vegetable glycerin by itself, which was purchased from Freedom Smoke USA as a DIY product, was cytotoxic (figure 2L). Only one refill fluid that contained glycerin, #6 Valencia, produced an aerosol that was not cytotoxic (figure 2F).

Solvent cytotoxicity

In figure 4, solvent aerosols and aerosols from refill fluids with known solvent compositions read off the label were plotted based on IC50 values and solvent composition. Generally, refill fluid aerosols containing glycerin were the most cytotoxic. In contrast, aerosols containing a mixture of propylene glycol and glycerin were overall less cytotoxic (figure 4). This pattern was also seen for the authentic aerosolised solvents, where propylene glycol showed lower cytotoxicity than glycerin.

Figure 4

Relationship between solvent and aerosol cytotoxicity. The IC50s (concentration in percent) of the aerosols for the hPF are plotted for each product denoted by inventory number. Lower IC50 values indicate higher cytotoxicity. Points plotted at 2.5 were not potent in the MTT (3-(4,5-dimethylthiazol-2-yl)−2,5-diphenyltetrazolium bromide) assay, as an IC50 value could not be generated. The ‘PG or VG only’ column denotes aerosols made from PG and VG purchased from Freedom Smoke USA. The ‘VG refill fluids’ column denotes aerosols made from refill fluids containing glycerin, and the ‘PG + VG refill fluids’ column denotes refill fluids that contain propylene glycol and glycerin. Each refill fluid was categorised based on the manufacturer’s labelling. hPF, human pulmonary fibroblast; IC50, inhibitory concentration at 50%.


This study compared the cytotoxicity of refill fluids and their corresponding aerosols, established a hierarchy of potency for the e-cigarette aerosols, compared the sensitivity of three different cell types to e-cigarette aerosols and demonstrated that aerosols made from cartomizers with glycerin were more cytotoxic than those made from propylene glycol.

Approximately 74% of the aerosols and their corresponding refill fluids had similar cytotoxicity classification. This correlation suggests that using refill fluids for screening purposes may be useful, practical and economical since there are thousands of unique refill fluids available, which may preclude testing large numbers of aerosols.1 Developing a screening method in which refill fluids are first tested for toxicity followed by testing aerosols from those fluids that show toxicity would be less expensive, faster and less labour intense than producing aerosols for every sample. However, some toxic aerosols may be missed with this strategy and when testing for chemical modifications or generation of new chemicals due to heating, screening of the aerosols would be necessary. In cases where the refill fluid contains only glycerin, predicting aerosol toxicity from refill fluid data would be over 90% accurate based on our data.

Other studies on the cytotoxicity of e-cigarette aerosols have found variable results. In concordance with our study, several groups have reported e-cigarette toxicity and activation of cellular stress pathways, including oxidative stress and inflammation.8 9 11 12 39–42 Similar to our results, one group found that aerosol toxicity varied when using different refill fluids with the same e-cigarette device and therefore concluded that the fluid composition is an important source of toxicity.43 In contrast to our study, some groups have concluded that e-cigarette refill fluids and aerosols are not cytotoxic or have relatively low cytotoxicities.44–47 These differences in cytotoxicity that have been reported by different laboratories could be due to variables such as cell type, species, the brand of e-cigarette studied, the voltage/wattage and the method of aerosol collection. Other groups have used air–liquid interface systems to directly expose human lung cells to e-cigarette aerosols. These studies have shown that a variety of toxicological effects arise including oxidation, stimulated inflammatory response, decreased cell viability, decreased metabolic activity and alterations in gene expression when using a 3D system.7 39 48–50 As a follow-up to our current screen, we are testing a subset of our refill fluids using an air–liquid interface system with intermittent exposure to determine if toxicity is similar in 3D culture.

Several studies have identified flavouring chemicals in e-cigarettes that may be potentially harmful including cinnamaldehyde (cinnamon), benzaldehyde (cherry) and 2,5-dimethypyrazine (chocolate).28 51 52 The hierarchy of potency generated for the aerosol dose–response data in our study shows that a variety of flavoured refill fluids are cytotoxic. Of the six most cytotoxic aerosols, five were in the creamy/buttery flavour category exclusively (ie, #19 Butterfinger, #28 Caramel, #30 Butterscotch, #29 Butterscotch and #26 Caramel), while the sixth was a mix of tobacco, chocolate and creamy/buttery (ie, #4 Swiss Dark). These six products are glycerin based and have IC50 values lower than that of the vegetable glycerin DIY product (#33). Due to this, it is plausible that their increased potency is related to the flavouring chemicals. Diacetyl and 2,3-pentanedione, which have been linked to lung disease in humans and rats,53–55 are frequently found in creamy/buttery refill fluids,20 56 and could be contributing to the cytotoxicity observed in our study.

All aerosols tested in this study, including those with relatively low potency, were significantly different from untreated controls except for #16 Summer Peach (figure 1B), and the majority of the refill fluids produced aerosol that reduced cell survival to 70% of the control or lower. While our in vitro studies cannot be applied directly to humans, some human data have shown that chronic exposure to e-cigarette aerosol may be linked to adverse health effects.6 13 14 However, it is probable that not all users are equally affected by e-cigarette products, as several studies have reported beneficial effects or no major adverse effects.6 13 14 57–59 hESCs, which model a postimplantation stage in human development,33 were often more sensitive to e-cigarette aerosols than differentiated adult cells. This trend is in line with the our original fluid screen, in which the two stem cell types were more sensitive than adult lung cells.27 Studies that have addressed e-cigarette use during pregnancy in animals have found alterations in cardiac development (zebrafish), behavioural changes (male mice) and impairment of developmental rate and brood size (Caenorhabditis elegans).42 60 61 Additionally, one case report describes a 1-day old infant with colonic necrotising enterocolitits, where the mother’s e-cigarette use throughout pregnancy and during labour was speculated to be the cause.62 The above studies indicate a need for additional research on the health effects of prenatal exposure to e-cigarette aerosol, especially since pregnant women sometimes switch from tobacco cigarettes to e-cigarettes to reduce harm to their embryo/fetus.63 64

One of the most important findings in this study was that aerosolised glycerin was more cytotoxic than aerosolised propylene glycol under normal conditions of use and without dry puffing. Our results are supported by other e-cigarette studies that used different cell types and showed that the solvents themselves caused toxicological effects in vitro.49 50 While the reported concentrations of carbonyl compounds in e-cigarette aerosols have varied,22 24 65–67 recent studies indicate that aerosolisation of glycerin produces carbonyls that could account for the cytotoxicity we observed.22 67 The cytotoxicity of the solvents is of particular importance given the growing popularity of glycerin-based refill fluids which produce a thicker aerosol cloud.14 Additionally, some refill fluids are advertised as propylene glycol free, to provide a product to users with propylene glycol allergies, and on many websites users are able to customise their propylene glycol/glycerin concentrations.14 68

In conclusion, this study shows that (1) aerosolised e-cigarette products vary in their cytotoxicity with creamy/buttery flavours being the most potent; (2) hESCs were often more sensitive than the adult lung cells to e-cigarette aerosols; (3) four patterns of cytotoxicity emerged when screening refill fluids and their corresponding aerosols; (4) about 74% of the time, cytotoxicity data from the refill fluids and aerosols correlated well; (5) using refill fluids to screen followed by focused screening of aerosols is a fast, inexpensive and less laborious method for identifying products that may produce cytotoxicity; (6) 91% of the time, glycerin-based refill fluids produced aerosols that were cytotoxic using a 2.85 V cartomizer style e-cigarette; and (7) when compared with propylene glycol and mixed solvent products, aerosols from glycerin-based products produced greater cytotoxicity. These data demonstrate that e-cigarette fluids and aerosols are often cytotoxic to human cells and may be hazardous to the respiratory system. This study provides new information to help guide the regulation of these products and to help users of e-cigarettes avoid products that may be harmful to their health.

What this paper adds

  • Thousands of refill fluids with unique flavourings are available online and in shops around the world making screening in aerosol form difficult.

  • Cytotoxicity in general did not change when refill fluids were stored at 4°C for 1 year.

  • Stem cells were generally more sensitive than lung fibroblasts and epithelial cells to aerosols.

  • Seventy-four percent of the time refill fluid cytotoxicity data accurately predicted the toxicity of the corresponding aerosol, signifying that toxicity screens could be done with fluids.

  • Aerosols from the creamy/buttery flavoured refill fluids were more cytotoxic than any other flavour group.

  • Glycerin-based refill fluids produced aerosols that were cytotoxic 91% of the time indicating that glycerin alone may be more harmful than propylene glycol or mixed solvent products.


We gratefully acknowledge Victor Camberos, Michael Dang, Jisoo Kim, Alex Razo and Eriel Datuin for their help making aerosol preparations, and My Hua for her helpful suggestions on the manuscript.



  • Contributors Conception and design: PT and RZB. Data collection and interpretation: RZB, YW and PT. Data analysis and writing of the manuscript: RZB and PT. Data processing and sample preparation: RZB and YW.

  • Funding Research reported in this publication was supported by grant number R21DA037365 to PT from the National Institutes of Health and the FDA Center for Tobacco Products. Rachel Behar was supported by an NIH NRSA Individual Predoctoral Fellowship (F31HL116121). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the Food and Drug Administration.

  • Competing interests None declared.

  • Patient consent The project did not involve humans.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All of the relevant data are included in the manuscript.