Toxicology of flavoring- and cannabis-containing e-liquids used

in electronic delivery systems

Aleksandr B. Stefaniak

a,*

, Ryan F. LeBouf

a

, Anand C. Ranpara

a

, Stephen S. Leonard

b

a

Respiratory Health Division, National Institute for Occupational Safety and Health, 1095

Willowdale Road, Morgantown, WV 26505, USA

b

Health Effects Laboratory Division, National Institute for Occupational Safety and Health, 1095

*

Declaration

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the

National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention. Mention of any company or

product does not constitute endorsement by the U.S. Government, National Institute for Occupational Safety and Health, or Centers

for Disease Control and Prevention.

Pharmacol Ther

. 2021 August ; 224: 107838. doi:10.1016/j.pharmthera.2021.107838.

Keywords

9

-THC); Toxicity

1. Summary

Electronic nicotine delivery systems such as electronic cigarettes (e-cigarettes) are used to

heat an e-liquid composed of humectants and sometimes flavorings and nicotine. The heated

cannabis as dried plant material or a concentrated extract, either by itself or dissolved in an

e-liquid. Since the introduction of e-cigarettes in the United States in 2007, these devices

rapidly gained popularity, especially amongst U.S. youth, and by 2014, were the most

popular tobacco product for this demographic, overtaking use of regular tobacco cigarettes.

The rise in popularity of e-cigarettes has raised important public health questions, including:

1) given the attraction of flavorings on e-cigarette use, what is our understanding of the

toxicity of flavored e-liquids?, and 2) given the ease in which e-cigarettes can be used to

aerosolize substances, what is our understanding of the toxicity of substances underlying the

outbreak of lung injury related to consumption of cannabis and nicotine products that

occurred predominantly in the United States termed e-cigarette, or vaping, product use-

flavorings in e-liquids and whether e-cigarettes pose a risk for dependence or addiction to

nicotine for a new generation of youth. To better understand the state of knowledge on the

toxicity of flavored e-liquids, we reviewed literature in the PubMed and Scopus databases

and identified 67 original articles that evaluated toxicity of flavored e-liquids using cellular

in vitro

, rodent

in vivo

, and human models. Whether as bulk liquid or aerosol from an e-

cigarette, some flavored e-liquids induced toxicity in the respiratory tract (cytotoxicity,

generation of reactive oxygen species, and impairment of clearance mechanisms),

cardiovascular and circulatory systems (impaired nitric oxide (NO) signaling and other

effects related to endothelial dysfunction), skeletal system (altered gene expression in

osteoblasts, toxicity in the oral cavity), and skin (cytotoxicity). In general, embryonic cells

liquids were immune sensitizers, irritants, or genotoxic. At least 65 individual flavoring

ingredients in flavored e-liquids were observed to contribute to reported toxic effects.

Cinnamaldehyde was most frequently reported to be cytotoxic, followed by vanillin,

menthol, ethyl maltol, ethyl vanillin, benzaldehyde and linalool. With regards to our second

question, among EVALI patients, 82% reported using a delta-9-tetrahydrocannabinol (Δ

THC, the main psychoactive component in cannabis) containing e-cigarette, or vaping,

product, 33% reported only using a Δ

-THC-containing product, 57% reported using any

nicotine-containing product, and 14% used only a nicotine-containing product. Our literature

search identified 33 articles of interest related to EVALI. Five of the 33 articles of interest on

EVALI contained information on

in vitro

or

in vivo

pulmonary toxicity. Vitamin E acetate

liquids (composition, bulk or vaped form), modeled systems (cell type, culture type, and

dosimetry metrics), biological monitoring, secondhand exposures and contact with residues

that contain nicotine and flavorings, and causative agents and mechanisms of EVALI

toxicity.

2. Introduction

Electronic delivery systems are devices that heat a liquid solution (e-liquid) or dry material

to volatilize its constituents, which are inhaled by the user in the form of an aerosol. Devices

intended for nicotine delivery are referred to as electronic nicotine delivery systems (ENDS)

and include electronic cigarettes (e-cigarettes or e-cigs), e-cigars, e-pipes, and e-hookahs.

Early versions of e-cigarettes were intended to mimic the tobacco smoking experience but

experience is termed “vaping.” In 2016, the U.S. Food and Drug Administration (FDA),

under authority of the Family Smoking Prevention and Tobacco Control Act, began to

regulate e-cigarettes as tobacco products, which includes the use of flavors in products.

Many flavorings used in e-liquids for e-cigarettes fall under the “generally recognized as

safe” (GRAS) provision in the Federal Food, Drug, and Cosmetic Act under the jurisdiction

of the FDA. Under FDA regulation, any substance that will be added to food is subject to

premarket approval, unless it is generally recognized, by scientific experts, as safe under the

conditions of its intended use. FDA determines the safety of the substance if it is subject to

premarket approval whereas qualified experts outside of government can submit a GRAS

notification to the FDA for their approval. The Flavor and Extract Manufacturers

of flavorings to be used in foods (as defined in Section 201(f) of the Act). Flavor and Extract

Manufacturers Association nominations of flavorings for GRAS status only assesses safety

for exposure through ingestion. Approval of their nominations by FDA does not provide

regulatory authority for the use of a flavoring in e-liquids where exposure is via inhalation

from vaping. The use of flavorings with GRAS status in e-liquids has raised concern by

public health experts and the Flavor and Extract Manufacturers Association alike of possible

toxicity (FEMA, 2020). Herein, when discussing e-liquids, the terms “flavor” or “flavored”

refer to a taste sensation (e.g., fruity) of an e-liquid and the term “flavoring” refers to the

specific chemical that imparts a taste (e.g.,

α

-ionone is a flavoring for raspberry). When

discussing flavored e-liquids, general flavor categories are in lowercase (e.g., fruity) whereas

plant contains less 0.3% of Δ

9

-THC. Though hemp has lower Δ

9

have a higher concentration of cannabidiol (CBD). Vaporizers include table top devices to

heat dried cannabis without combustion at moderate temperatures to create an aerosol that is

inhaled and or portable, pocket pen-vaporizers to heat its wax and oil extracts at higher

temperatures to deliver aerosolized cannabinoids in a form that can be inhaled (Giroud et al.,

2015; Varlet et al., 2016). In addition to electronic products specifically designed for

cannabis, other devices, including e-cigarettes that are designed to deliver nicotine, can be

modified to deliver various substances. For example, e-cigarettes can be used after-market to

inhale alcohol, cannabis, amphetamines, cocaine, and heroin (Breitbarth, Morgan, & Jones,

cells and discussed potential biomarkers (Kaur, Muthumalage, & Rahman, 2018). The

current review article extends the work of Kaur et al. to include more recent literature on

lung toxicity and provides a general overview of all organs and systems currently known to

be impacted by flavored e-liquids. For a review of the clinical aspects of EVALI, see Cherian

et al. (Cherian, Kumar, & Estrada, 2020).

Second generation e-cigarettes

were typically larger than regular tobacco

cigarettes, had medium-sized rechargeable batteries, evolved to contain a

powerful atomizer to deliver greater energy that enhanced nicotine delivery, and

large refillable cartridges for e-liquids (Schmidt, 2020). Variations of atomizers

in second (and third) generation e-cigarettes included “cartomizers,” which were

similar in design to atomizers but utilized a synthetic filler material that was

wrapped around the heating coil to absorb e-liquid, and “clearomizers”

composed of a clear tank but no filler material. Some second and future

generation e-cigarettes had a manual switch that allowed the user to modulate

referred to as “personal vaporizers” (Protano et al., 2018). Other designs allowed

the user to automatically puff by inhaling through the e-cigarette mouthpiece

without the need to depress a switch and had adjustable voltages.

material (Breitbarth et al., 2018;Varlet, 2016; Varlet et al., 2016). Ramamurthi et al.

provided an insightful review of stealth vaping of cannabis using portable electronic devices

and pointed to the commercial availability of products designed to hide the device by taking

on the appearance of common items such as an ink pen, travel coffee mug, electronic car key

fob, small electronics such as a remote control or an iPod®, a small mobile phone, a candy

dispenser (vaporizer placed inside a package of Tic Tac® breath mints), and integrated into

clothing such as hooded sweatshirts and backpacks (Ramamurthi et al., 2019).

(Breitbarth et al., 2018; Giroud et al., 2015).

proportion in an e-liquid, it contributes to an experience referred to as “throat hit”, which is a

sensation produced at the back of a user’s throat upon inhalation of nicotine that may range

from pleasant to harsh (Smith, Heckman, Wahlquist, Cummings, and Carpenter, 2020). VG

is used at a higher proportion than PG if a user seeks a denser exhaled cloud and is popular

among “power vapers” or “cloud chasers” who perform tricks such as creation of exhaled

shapes (Schmidt, 2020).

E-liquids contain nicotine in free-base (basic pH ~8 to 10) or salt (acidic pH) form (El-

Hellani et al., 2015). The form of nicotine in an e-liquid and resultant aerosol influences its

bioavailability, which has varied with e-cigarette generation. E-liquids used in many third-

generation and prior e-cigarettes contained 18 to 95% of their total nicotine in free-base

and the ratio of nicotyrine to nicotine in the e-liquid and aerosol generated by an e-cigarette

increased over time. Nicotyrine is potentially toxic but also inhibits the

of nicotine, hence, it is hypothesized to be a potentially useful smoking cessation aid

(Martinez, Dhawan, Sumner, & Williams, 2015).

variety of flavorings that are a major selling point of all ENDS (Berg, 2016). The presence

of flavorings may add to the addictive effects of e-cigarettes (Soule, Lopez, Guy, & Cobb,

prominent flavors were fruit (34%), tobacco (16%), and dessert (10%) (Havermans et al.,

2021). At least 210 different flavorings chemicals were used to create these flavored e-

liquids and the mean number of flavorings per e-liquid was 10 (Krusemann et al., 2021).

Table 1 is a list of 65 flavoring chemicals present in flavored e-liquids that have been

reported to induce

in vitro

a subset of all flavorings that were tested in the studies listed in Table 1; some flavorings did

not induce toxicity in some of the reviewed studies.

E-liquids for pod mods contain humectants, water, and usually flavorings but differ from e-

liquids used in previous generation e-cigarettes in two important ways that maximize

nicotine uptake to blood: 1) they contain an acid additive, and 2) nicotine in this matrix is in

the form of a protonated salt (Gotts, 2019; Harvanko, Havel, Jacob, & Benowitz, 2020;

Jackler & Ramamurthi, 2019; Ramamurthi et al., 2019; Talih et al., 2019). A regular tobacco

cigarette contains approximately 1.5–2% nicotine which is equivalent to 1.5–2 mg

nicotine/mL by volume. JUUL® brand is the most popular pod mod style device in the

United States and in 2017, accounted for nearly 40% of all e-cigarette sales and over 70% of

the salt form that has a lower pH (~4.9) than free-base nicotine (~8 to 10), thereby allowing

high levels of nicotine to be inhaled (and absorbed into blood) more easily and with less

irritation or harsh throat hit compared with free-base nicotine in regular tobacco cigarettes

and earlier generation e-cigarettes (Gotts, 2019; Harvanko et al., 2020; Jackler &

Ramamurthi, 2019; Schmidt, 2020; Talih et al., 2019). JUUL® brand flavored e-liquids once

included Cool Mint, Classic Menthol, Mango, Fruit Medley, Cool Cucumber, Crème Brulee,

Classic Tobacco, and Virginia Tobacco. As of November 2019, JUUL® only sells menthol

and tobacco flavored e-liquids. Note that other manufacturers have developed flavor

enhancement pods that attach to the mouthpiece of JUUL® and other brand pod mod

devices to mix flavorings with the user’s nicotine salt e-liquid (Cwalina, Leventhal, &

4.2. E-liquids for cannabis delivery

E-cigarettes are used to vape Δ

laboratory) dispersed in e-liquids. Δ

9

-THC extracts, because of their unique physiochemical

properties, are difficult to disperse in PG/VG humectants. Δ

9

highly viscous, semi-solid materials that are usually mixed with diluents, which might

include vitamin-E acetate (VEA), medium chain triglycerides, coconut oil, squalane, or

terpenes to form an e-liquid (Blount et al., 2020; Chand, Muthumalage, Maziak, & Rahman,

9

al., 2020; Duffy et al., 2020). Some e-liquids for Δ

9

-THC extracts were reported to contain

pure PG as a diluent (Giroud et al., 2015; Peace, Stone, Poklis, Turner, & Poklis, 2016;

Varlet et al., 2016). When heated at temperatures used to vape cannabis oils, PG can form

acetaldehyde and formaldehyde (Troutt & DiDonato, 2017). Polar synthetic cannabinoids

readily dissolve in the same e-liquid formulations that are used for nicotine delivery

(Apirakkan et al., 2020; Breitbarth et al., 2018). Similarly, CBD can be dispersed in the

same e-liquid formulations used for nicotine delivery (Grafinger, Kronert, Broillet, &

Weinmann, 2020; Peace, Butler, Wolf, Poklis, & Poklis, 2016).

United States, 19.6% of high school students and 4.7% of middle school students were

current e-cigarettes users (Wang et al., 2020). Globally, the use of ENDS is one of the most

popular ways to inhale cannabis (Breitbarth et al., 2018).

cytotoxicity in adjacent cell culture wells. Next the authors analyzed the chemical

composition of e-liquids that exhibited cytotoxicity and identified four common flavorings:

cinnamaldehyde, 2-methoxycinnamaldehyde, dipropylene glycol, and vanillin. HPF cells

were exposed to authentic standards of each flavoring and all were cytotoxic;

cinnamaldehyde and 2-methoxycinnamaldehyde were the most potent (Behar et al., 2014).

Results of this publication sparked a debate with the Farsalinos laboratory on whether it was

appropriate to test diluted e-liquids since, when aerosolized, they are heated and the

characteristics of the aerosol might differ from the bulk liquid (Behar, Davis, Bahl, Lin, &

Talbot, 2014; Farsalinos, Romagna, & Voudris, 2014). The Talbot laboratory exposed A549

and HPF cells to aerosolized Cinnamon Ceylon flavored e-liquid and aerosolized

Talbot laboratory reaffirmed that both the e-cigarette generation and applied voltage

influenced aerosol production (including the formation of new substances), which in turn

affected cytotoxicity (Behar, Luo, McWhirter, Pankow, & Talbot, 2018) and that observed

cytotoxicity from exposure to diluted e-liquids and aerosolized e-liquids agreed 74% of the

time, which indicated that bulk liquids have utility to screen for cytotoxicity (Behar, Wang,

& Talbot, 2018). Other researchers, including Otreba et al. have independently confirmed

that cytotoxicity of aerosolized flavored e-liquids increased with applied e-cigarette voltage

(Otreba, Kosmider, Knysak, Warncke, & Sobczak, 2018).

butterscotch flavored e-liquids were among the most cytotoxic (Singh, Luquet, Smith,

Potgieter, & Ragazzon, 2016).

pulmonary microvascular endothelial cells). Both aerosols were cytotoxic in the mono- and

co-culture systems, though the monoculture was generally more sensitive to cytotoxic effects

SA-β-gal is indicative of cellular senescence (alterations in cellular homeostasis consistent

with pre-mature aging). TGF-β1 controls differentiation of fibroblast cells into

myofibroblasts, and inhibition of this growth factor indicated compromised wound healing

responses in cells.

epithelial cell lines (Beas-2B, IB3-1, C38, and CALU-3), one mouse macrophage cell line

(J774), one human monocyte cell line (THP-1), and one human fibroblast cell line (Wi-38).

These cell lines were selected to test the effects of aerosolized flavored e-liquids on multiple

respiratory cell types that would encounter inhaled e-cigarette aerosol: bronchial epithelial

cells that line the upper respiratory tract, underlying fibroblast cells, and macrophages,

which are immune cells that function to remove foreign material from lung surfaces. As

expected, different cell types exhibited different sensitivity to aerosolized flavored e-liquids.

In general, Beas-2B lung epithelial cells were most sensitive and aerosolized Strawberry and

prepared e-liquids and HBE cells. The authors reported that individually 2-acetylthiazole,

allyl hexanoate,

-pinene, citronellol, guaiacol, linalool, methyl anthranilate, 3-methyl-2,4-

nonanedione, 3-(methylthio) propionaldehyde, and phenethyl alcohol e-liquids exhibited

increased cytotoxicity; citronellol and

α

-pinene were the most cytotoxic. When they

evaluated the cytotoxicity of flavoring mixtures, the cytotoxicity of mixtures differed from

that of the individual flavorings and citronellol was the main driver of toxicity while other

flavorings contributed to synergistic effects (Marescotti et al., 2020). Tissue models better

mimic

in vivo

conditions than submerged monocultures because they contain differentiated

cell types that are present in the respiratory epithelium. Using ALI systems, aerosolized

Blueberry flavored e-liquid was not cytotoxic in the EpiAirway

3D tissue model (Czekala

Pumpkin Pie, Chocolate Banana, Cherry Kola, Kola, Hot Cinnamon Candies, Mojito, Green

Gummies, Vanilla Bean, and Menthol Tobacco flavored e-liquids were most cytotoxic to

HEK-293T cells. It is important to note that, though often used in toxicology studies because

of their robustness, HEK are human kidney epithelial cells, not respiratory cells. The authors

further evaluated a subset of 14 flavored e-liquids and reported that they were cytotoxic in

A549 human lung epithelial, HBE, and primary alveolar macrophage cells. Their data

revealed a weak correlation between the presence or absence of flavorings in e-liquids and

cytotoxicity; however, there was a correlation between cytotoxicity and the concentration of

vanillin and cinnamaldehyde flavorings in e-liquids and vanillin was identified as a major

computationally derived scoring system for each tier to create a single summary score of all

observed toxic effects of an e-liquid. Their scoring system was applied to 28 flavoring

chemicals alone or in mixtures, and as noted earlier in this section, cytotoxicity of mixtures

differed from that of the individual flavoring constituents (Marescotti et al., 2020). Based, in

part, on the results of these studies, the tobacco cigarette industry touted their systems

toxicology approach as a valuable tool to screen single flavoring substances and rank them

based on their toxicity “so that manufacturers can develop and/or produce e-liquids with

nontoxic flavor composition and doses” (Marescotti et al., 2020). As noted in Section 5.1,

use of flavorings in e-liquids is a known major attractant for youth to begin e-cigarette use.

weakly correlated with menthol and vanillin flavoring concentrations. The authors noted that

the U.S. FDA has raised concerns that JUUL® use may pose risk of addiction to nicotine for

a new generation of adolescents and serve as a gateway to use of regular tobacco cigarettes.

They also noted that their data raised a new concern that the high levels of flavorings in

JUUL® e-liquids can damage or kill lung cells (Omaiye, McWhirter, Luo, Pankow, &

Talbot, 2019). As noted in Section 4.1, as of November 2019, JUUL® only sells Menthol,

Classic Tobacco, and Virginia Tobacco flavored e-liquids. Recently, Lamb et al. evaluated

the effects of aerosolized JUUL® brand Menthol and Virginia Tobacco e-liquids on

mitochondrial function. They reported that aerosolized Menthol flavored e-liquid caused

Secondhand exposure to regular tobacco smoke is associated with development of otitis

more cytotoxic compared with Tobacco flavored e-liquid. The authors concluded that

flavored e-liquids were cytotoxic and could cause otitis media in middle ear epithelial cells

via reduced cell viability and stimulation of inflammatory cytokines and mucin production.

cigarette, so the only secondhand exposure potential is the aerosol that is exhaled by a user.

The composition of mainstream aerosol (what is inhaled by the user) differs from that of the

antioxidant defenses (Ho, Magnenat, Gargano, & Cao, 1998) and if cells are unable to

maintain this redox balance, it may result in a chronic inflammatory state in the respiratory

system. This inflammatory state may result in damage to the cells involved and to the

surrounding tissue via activation of signaling pathways, inflammatory cytokine production,

altered gene expression, and other cellular modifications.

Several research groups have screened flavored e-liquids for capacity to produce ROS using

cell-free and cellular systems. Lerner et al. evaluated the capacity of 22 flavored e-liquids to

generate ROS using a cell-free fluorescent probe. In this study, aerosol was generated using

an e-cigarette attached to a smoking machine; the aerosol was passed through a bubbler that

contained 2’,7’dichlorofluorescein dye solution then analyzed for oxidized

flavored e-liquids, which indicated that ROS may be inhaled directly into the lung during e-

cigarette use. Further, generation of ROS in aerosolized e-liquids was impacted by the age of

the e-cigarette heating coil (see e-cigarette schematics in Fig. 1): more ROS were produced

from a new coil compared with coils that were previously used at least 50 times (Lerner et

al., 2015). Muthumalage et al. also evaluated cell-free ROS generation from flavored e-

liquids aerosolized with new and aged heating coils using the 2’,7’dichlorofluorescein dye

method. When aerosolized with a new atomizer, American Tobacco, Mystery Mix, and

Mixed Flavors e-liquids generated ROS and when aerosolized with a used atomizer, Café

Latte, Cinnamon Roll, and Cotton Candy flavored e-liquids generated ROS. In this study,

hydrogen peroxide (H

and used heating coils were similar (Muthumalage et al., 2018). Unfortunately, the same

flavored e-liquids were not tested with both conditions of coils (new or used), which

precluded inference as to the influence of coil age on ROS generation. Zhao et al.

characterized cell-free ROS generation from two flavored e-liquids. These authors used

Trolox, a water-soluble form of vitamin E to measure ROS production. Trolox is oxidized to

quantified as the difference the samples with and without horseradish peroxidase. Zhao et al.

reported that aerosolized Fruit flavored e-liquid generated three times more total ROS and

H

compared with aerosolized Tobacco flavor e-liquid. Based on these and other

experimental results presented, the authors concluded that aerosolized e-liquids may contain

ROS precursors that can generate more ROS upon deposition in the lung and phagocytosis

by macrophages (J. Zhao et al., 2018). Iskander et al. reported no difference in oxidative

stress from exposure to an aerosolized flavored (not specified) e-liquid compared with an

unflavored e-liquid using a submerged monoculture of HBE cells and an ALI system with

SmallAir

™

(human small airway) 3D tissue model (Iskandar, Zanetti, Marescotti, et al.,

2019). Bitzer et al. screened 49 laboratory-prepared flavored e-liquids for oxidative capacity

and reported that a vanilla flavored e-liquid generated less ROS compared with an

unflavored PG/VG humectant mixture in a cell-free system (collected in alpha phenyl-N-tert

concentrations of total metals and generated more ROS in HBE cells compared with

aerosolized JUUL® Fruit Medley e-liquid. At the highest dose tested (25 puffs), all

aerosolized e-liquids caused oxidative stress (measured as reduction in HBE cellular

glutathione levels). Results were not compared with aerosolized humectant-only e-liquid

exposures nor aerosolized nicotine-free e-liquid exposures, as such, it is difficult to

disentangle the relative influence of flavorings, metals, and nicotine on reported results from

this study.

dissolved in PG/VG and aerosolized using an e-cigarette. Ethyl vanillin PG acetal, ethyl

vanillin, β-damascone, and δ-tetradecalactone were negatively correlated with ROS levels,

which suggested an inhibitive effect. Six flavorings (γ-decalactone, citral, ethyl maltol,

piperonal,

d

-limonene, and linalool) were positively correlated with ROS levels, which

indicated that they contributed to increased radical production. Further, linalool, piperonal,

and citral caused significant increases of lipid peroxidation products (Bitzer et al., 2018).

Muthumalage et al. measured ROS generation from seven flavoring chemicals using a cell-

free system (2’,7’dichlorofluorescein dye method). Diacetyl, cinnamaldehyde, maltol, o-

vanillin, and coumarin flavorings generated ROS in a concentration-dependent manner

whereas acetoin and 2,3-pentanedione only generated ROS at highest tested concentration

concentrations, and isoamyl acetate did not affect the response (Hickman, Herrera, &

Jaspers, 2019). As described in Section 7.1.1, the tobacco cigarette industry has proposed a

three-tiered systems toxicology approach to evaluation of e-liquids. The second step of their

approach was evaluation of the mechanisms of toxicity of e-liquids. Marescotti et al.

evaluated the impact of 28 flavoring chemicals, singly or in mixtures, using laboratory

prepared e-liquids and human bronchial epithelial cells. Of the 28 flavorings evaluated

singly, two, citronellol and

α

-pinene were reported to induce increased oxidative stress

(Marescotti et al., 2020).

2

™

pathogenesis of periodontal disease. In the only study to evaluate the oxidative capacity of

aerosolized flavored e-liquids in humans, Menicagli et al. measured salivary

malondialdehyde as a marker of lipid peroxidation. The authors reported that levels of

salivary malonidialdehyde were significantly higher in e-cigarette users who aerosolized a

tobacco-flavored e-liquid without nicotine, which suggested oxidative damage (Menicagli,

Marotta, and Serra, 2020). Collectively, the results of most studies indicate that flavored e-

liquid aerosol can disrupt cell function via oxidative stress pathways, which may result in

oral diseases.

flavored e-liquids compared with other tobacco product extracts and that IL-8 was released

by cells exposed to aerosolized flavored e-liquids (Misra et al., 2014). Clapp et al. reported

that Sini-cide (cinnamon flavored) e-liquid significantly suppressed IL-8 secretion by human

alveolar macrophage cells (Clapp et al., 2017). In another study, 36 flavored e-liquids and

seven flavoring chemicals were evaluated, and it was reported that specific flavored e-liquids

and acetoin flavoring suppressed IL-8 secretion by human pleura/pleural lymphocyte (U937)

cells (Muthumalage et al., 2018). Czekala et al. did not observe any change in IL-8 levels in

EpiAirway

flavored e-liquid compared with an unflavored e-liquid (Czekala et al., 2019). In contrast, it

Collectively, the available body of literature has demonstrated that ROS generation and

secretion of proinflammatory signaling molecules following exposure to flavored e-liquids

and flavoring chemicals is substance-specific and may be suppressed, unchanged, or

increased depending upon the e-liquid composition or flavoring chemical. Additionally, the

same flavoring did not induce cytokine secretion in different cell types, which indicated that

the choice of cell line is an important consideration in study design.

movement of foreign material that is deposited in the mucous layer that covers the

respiratory tract epithelium via the beating of cilia. This mechanism is sometimes referred to

as the “mucociliary escalator.” The ciliated portion of the respiratory tract extends from the

nose through the tracheobronchial region but excludes the alveolar region. Each cilia of the

respiratory tract beats in a coordinated fashion at the same frequency but in a phase-shifted

manner with its neighbors, which has the net effect of generating a wave that travels across

the epithelium and propels the overlying mucus layer (Bustamante-Marin & Ostrowski,

2017). In the nose, the mucus layer is propelled by ciliary action toward the pharynx

whereas in the tracheobronchial region the mucus layer is propelled by cilium toward the

larynx, where it is swallowed and excreted via the gastrointestinal tract. Mucociliary

™

Blueberry flavored e-liquid and aerosolized unflavored e-liquid using an ALI system

observation, Iskander et al. reported that exposure to aerosolized flavored (unspecified) e-

liquid and aerosolized unflavored e-liquid did not impact cilia beating frequency in

SmallAir

™

(human small airway) and EpiOral

™

(human mucosal) 3D tissue models using

an ALI system (Iskandar et al., 2019).

Sherwood and Boitano evaluated the impact of 2,5-dimethylpyrazine flavoring on HBE cells

with a high-capacity real-time cell analysis approach. Cells were treated with non-cytotoxic

concentrations of 2,5-dimethylpyrazine, followed by exposure to either forskolin (to raise

cyclic adenosine monophosphate) or exogenous adenosine triphosphate (to raise intracellular

Ca

2+

concentration). Exposure to both compounds resulted in a concentration-dependent

reduction in physiological response, which indicated a change in signaling molecules

suppressed cilia beat frequency (Clapp et al., 2019). One research group took a

transcriptomic approach to evaluate the influence of flavorings on mucociliary clearance.

The authors exposed HBE cells to authentic standards of diacetyl or 2,3-pentanedione using

an ALI system and identified 163 and 568 differentially expressed genes, respectively. Of

these genes, 142 were common to both flavorings; expression of several genes significantly

downregulated production of cilia. Further, exposure to these flavoring chemicals

significantly decreased the number of ciliated cells, which indicated the potential to impair

mucociliary clearance (Park et al., 2019).

7.1.4. Impairment of cell-mediated clearance—Macrophages, neutrophils, and

Natural Killer (NK) cells form part of the body’s first line of defense against foreign

susceptible to infection and/or contribute to impaired resolution of inflammation (Clapp et

al., 2017). For example, airway macrophages and neutrophils from regular tobacco cigarette

smokers and persons with chronic obstructive pulmonary disease have impaired ability to

phagocytize bacteria, which can make them susceptible to development of pneumonia

(Hickman et al., 2019; Ween et al., 2017). In one study, Kola and Sini-cide flavored e-liquids

impaired macrophage phagocytic capacity (Clapp et al., 2017). Aerosolized Apple and Irish

Cream flavored e-liquids reduced macrophage (THP-1 human monocytes from peripheral

blood)-mediated phagocytosis of bacteria (Gómez et al., 2020; Ween et al., 2017), an effect

that was demonstrated to be related to reduced expression of the bacteria phagocytosis

recognition receptor (SR)-A1 on macrophage cell surfaces (Ween et al., 2017). Clapp et al.

phagocytosis in a dose-dependent manner (

(banana) had no effect (Hickman et al., 2019). Clapp et al. isolated NK cells from venous

blood and exposed them to flavored e-liquids and reported that Cinnamon flavored e-liquids

aerosolized e-liquids on innate immune cells in the lung, Werley et al. exposed female

Crl:CD(SD) rats for 90 days via nose-only inhalation to aerosol produced using a tobacco

cigarette industry prototype e-cigarette with a flavored e-liquid (specific flavor not

identified) and an unflavored e-liquid with the same composition. Rats in the high flavored

e-liquid exposure group had higher alkaline phosphatase, lactate dehydrogenase, and total

protein values compared with rats in the low- and mid- flavored e-liquid exposure groups,

which suggested general cytotoxicity to lung cells. Levels of these parameters in the high

flavored e-liquid exposure group did not differ from the high-exposure vehicle control

group. The authors reported at 28 and 90 days, the high flavored e-liquid exposure group had

higher proportions of neutrophils than the low- and mid-exposure groups; however, there

was no difference in total and differential cell counts of alveolar macrophages, neutrophils,

lymphocytes, eosinophils, or total cell count compared with control animals, which

suggested no impact on cellular clearance (Werley et al., 2016).

exposed to aerosol generated from an e-liquid without nicotine had abnormal levels of

neuroregulatory gene expression, which suggested disturbancesin central metabolic

regulation in offspring (Greene & Pisano, 2019). Whether the non-nicotine components

responsible for these reported developmental effects were from exposure to flavorings is

unknown. More information on the developmental toxicity of e-cigarettes in general can be

found in a recent review article (Greene & Pisano, 2019).

the mineralization of bone and the production of collagen type I, a major structural

component of bone extracellular matrix. Inhalation of regular tobacco smoke is a risk factor

for osteoporosis, and the Heggland laboratory questioned whether flavored e-liquids would

also impair bone health. In their first study, they treated human MG-63 and Saos-2

osteoblast-like cells with diluted flavored e-liquids with or without nicotine (Otero et al.,

2019). Exposure to e-liquids induced dose-dependent cytotoxicity independent of nicotine

flavored e-liquids to alter osteoblast gene function. Mango Blast flavored e-liquid with or

without nicotine significantly increased collagen type I protein expression. Next, the

Heggland laboratory evaluated in more detail the capacity of diluted and aerosolized

cinnamon flavored (Napalm, Cinn Candy) e-liquids to induce osteotoxicity (Wavreil &

Heggland, 2020). Exposure to aerosolized Napalm and Cinn Candy flavored e-liquids

induced significant cytotoxicity in MG-63 cells but did not alter collagen type I protein

expression. MG-63 cells exposed to cinnamon flavored e-liquids or aerosolized cinnamon

flavored e-liquids exhibited significantly increased ROS production. Collectively, results

from these studies indicated that coffee, fruit, menthol, and cinnamon flavored e-liquids

were cytotoxic and/or had capacity to alter Col1a1 gene expression in osteoblast-like cells.

The capacity of cinnamon flavored e-liquids to induce cytotoxicity may be related to

oxidative stress. In a study funded by the tobacco cigarette industry, Reumann et al. reported

Another potential impact of sweet flavored e-liquids on the skeletal system is their

cariogenic (cavity) potential in teeth. Kim et al. prepared e-liquids composed of humectants

and nicotine with ethyl butyrate (pineapple), ethyl maltol (cotton candy), hexyl acetate

(apple), and triacetin (velvety/smoky) flavorings or without flavoring chemicals.

Streptococcus mutans

donor teeth, which led to a two-fold increase in biofilm formation and up to a 27% decrease

in enamel hardness compared with aerosolized unflavored e-liquids. Additionally,

aerosolized flavored e-liquids that contained ethyl butyrate, hexyl acetate, and triacetin were

associated with bacteria-initiated demineralization of enamel and ethyl maltol inhibited

Streptococcus mutans

7.5. Allergenicity and irritation

The exact number of flavoring chemicals used in e-liquids is unknown, though one study

reported that at least 210 distinct flavorings were used to create over 16,000 flavored e-

liquids (Krusemann et al., 2021). Given the vast number of flavorings that are inhaled into

the body and the potential for dermal exposure among e-cigarette users and those who work

e-liquids to induce respiratory or skin sensitization (Stevenson et al., 2019). This testing

strategy consisted of assays that measured changes in the transcriptional profiles of 200

genomic biomarkers that were relevant to respiratory (type I immediate IgE-mediated

hypersensitivity) or dermal (type IV delayed cell-mediated type hypersensitivity)

sensitization and a predictive model to classify substances. Based on the GARD assays,

none of the flavored e-liquids were classified as respiratory sensitizers; however, both the

Blu Cherry and the unspecified flavored e-liquids were classified as weak dermal sensitizers

under European regulatory requirements. Some flavorings are known allergens; however, it

is largely unknown if these pure flavoring chemicals can induce sensitization as a

component of an e-liquid mixture (Stevenson et al., 2019). An important outcome of the

research by Stevenson et al. was the demonstration that the GARD assays, originally

developed to assess the sensitization potential of pure chemicals, showed promise to

Asthmatics who smoke regular tobacco cigarettes are susceptible to the effects of inhaled

substances during smoking and are known to have worse asthma control, need more

unscheduled healthcare visits, and require greater medication (Chapman et al., 2019).

Despite this association, little information exists on the effect of e-cigarette use on

asthmatics. To address this knowledge gap, Chapman et al. exposed male and female Balb/c

mice to house dust mite allergen and aerosolized flavored e-liquids over a period of 21 days.

The number of macrophages in bronchial alveolar lavage fluid (BALF) was elevated in mice

exposed to house dust mite and aerosolized Black Licorice flavored e-liquid (but not Kola,

Banana Pudding, or Cinnacide aerosolized e-liquids) without nicotine. Exposure to

aerosolized Black Licorice, Banana Pudding, and Cinnacide flavored e-liquids with 12

nicotine had significant but varied effects on features of allergic airways disease. This

conclusion was important given the appeal of flavored e-liquids to youth (see Section 5.1)

and the association between ever or current e-cigarette use and asthma in teenagers (Wills,

Choi, & Pagano, 2020).

Non-immunologic responses to flavored e-liquids or flavoring chemicals may include

respiratory irritation, which often serves as a sentinel warning of potential toxicity of an

inhaled chemical. Irritation is manifest as responses from chemical activation of

chemosensory receptors in airway-innervating nerves (Erythropel et al., 2019). The transient

TRPV1 receptors at lower concentrations. Based on these data, the authors concluded that 1)

aldehyde flavoring PG acetals are formed

in situ

in flavored e-liquids and persist when an e-

liquid is aerosolized; 2) these PG acetals have the capacity to induce a stronger irritation

response compared with the free parent aldehyde alone; 3) PG acetals may produce stronger

sensory irritation effects than the free aldehydes alone; and 4) that the toxicological

properties of PG acetals differed from the parent aldehyde flavorings and e-liquid

constituents. The authors advocated for a standard approach to detect and characterize newly

formed compounds in flavored e-liquids and their aerosol to assess toxicological effects

(Erythropel et al., 2019).

7.5. Genotoxicity

have evaluated the genotoxic potential of flavored e-liquids (Al-Saleh et al., 2020; Behar et

al., 2016; Misra et al., 2014; Menicagli et al., 2020; Welz et al., 2016). Misra et al. were the

first to evaluate genotoxicity and they used an

in vitro

measures chromosome damage based on the cytokinesis-block technique. Briefly, this assay

quantifies inhibition of actin filaments by cytochalasin B during cytokinesis and the

formation of “daughter” cells to distinguish between undivided cells and cells that

electrical field is applied to the chamber, intact cellular DNA remains stationary but

damaged DNA migrates in the gel, which results in a figure shaped like a comet (i.e.,

undamaged DNA in the head and damaged DNA forming the tail). The extent of migration

can be characterized as the percentage of DNA in the tail and olive tail moments (Welz et al.,

2016). The percentage of tail DNA in the gel is a measure of the relative fluorescent

intensity in the head compared with the tail. Olive tail moments represent the product of the

relative amount of DNA that migrated in the gel and the median migration distance. The

percentage DNA in the tail is a more robust indicator that enables inter-comparison of study

data, whereas olive tail moments may not be comparable between studies (Welz et al.,

2016). Results of Comet assays revealed that the percentage of cells with comet tails, comet

tail length, and olive tail moments were significantly increased from exposure to aerosolized

cinnamaldehyde flavoring at non-cytotoxic concentrations in HESC and HPF cells, but not

concluded that some flavored e-liquids may present risk for development of HNSCC (Welz

et al., 2016). In another study of HNSCC, Tsai et al exposed the Ca9-22 oral gingival

squamous carcinoma cell line and CAL-27 human tongue squamous cell line to Cinnamon

Red Hots (contains cinnamaldehyde) or Apple Juice flavored e-liquids with and without

nicotine (Tsai et al., 2020). For Ca9-22 oral gingival cells, exposure to Cinnamon Red Hots

flavored e-liquid increased invasiveness (measured as cell movement through a microporous

membrane toward a chemoattractant in a Boyden chamber) with or without nicotine whereas

exposure to Apple Juice flavored e-liquid decreased cell invasion independent of nicotine.

For CAL-27 tongue cells, Cinnamon Red Hots flavored e-liquid decreased cell invasion with

or without nicotine but for Apple Juice flavored e-liquid there was no difference with or

liquids and further potentiated by the presence of nicotine. For Ca9-22 oral gingival cells,

both flavored e-liquids increased secretion of IL-1α levels but only Apple Juice flavored e-

liquid increased IL-8 secretion; the presence of nicotine attenuated cytokine levels. For

CAL-27 cells, both flavored e-liquids increased secretion of IL-1α, but IL-8 levels were only

increased by Apple Juice flavored e-liquid.

Al-Saleh et al. evaluated 33 brands of flavored e-liquids, nearly all of which contained

quantifiable levels of menthol flavoring. Several brands of e-liquids that contained menthol

flavoring induced DNA damage as measured by tail movement (Comet assay) in CHO and

cells compared with air-exposed controls, which suggested a possible pathway for risk of

oral cancers (Menicagli et al., 2020).

whereas the other investigators exposed cells to diluted flavored e-liquids. The potential for

most cells to be directly exposed to bulk e-liquid is likely low (with the exceptions of

nearly two thirds were male (www.cdc.gov/EVALI). There were 2,022 hospitalized patients

who had data on substance use as of January 14, 2020; 82% reported using Δ

9

containing products, 33% reported exclusive use of Δ

reported using nicotine-containing products, and 14% reported exclusive use of nicotine-

containing products (www.cdc.gov/EVALI). Our literature searches for this review identified

33 peer-reviewed articles on EVALI for inclusion, though the vast majority (22/33) lacked

toxicological data. Six articles provided results of e-liquid or constituent characterization or

hypothesized an underlying toxicological mechanism for EVALI (Attfield et al., 2020;

Blount et al., 2020; Chand et al., 2019; Muthumalage et al., 2020; Narimani & da Silva,

2020; Wu & O’Shea, 2020). Four articles evaluated the toxicity of common cannabis extract

diluents on lung cells (Bhat et al., 2020; Jiang et al., 2020; Matsumoto et al., 2020;

Muthumalage et al., 2020) and one article reported an EVALI-like condition in rat lungs

from exposure to an e-liquid that did not contain Δ

9

-THC, VEA, or nicotine (Kleinman et

al., 2020). For a detailed review of

in vivo

studies of EVALI, the reader is referred to a

recent review article (Feldman, Stanton, & Suelzer, 2021).

8.1. Possible causative agents

Polyethylene glycol 400 and PG produced formaldehyde and/or acetaldehyde at higher

levels compared with VG and MCT (Troutt & DiDonato, 2017). None of these aldehydes

produce the pathology observed with EVALI, but the data provided valuable insights during

the outbreak because it suggested that other ingredients could be involved in the toxic

response. As part of the EVALI investigation, Blount et al. obtained BALF from 51 EVALI

patients in 16 states and measured the concentrations of several possible toxic substances,

9

product that contained Δ

and limonene were quantified in 94%, 2%, and 3% of EVALI patient BALF samples,

respectively but not in BALF samples from a control group. The authors postulated that the

general absence of other toxicants (plant oils, medium-chain triglyceride oil, coconut oil,

petroleum distillates, and terpenes) in BALF of EVALI patients discounted the role of these

constituents as a primary cause of EVALI.

could penetrate a layer of lung surfactant to align the molecule in parallel with

phospholipids, thereby interfering with surfactant function (Blount et al., 2020). Wu and

O’Shea reported that VEA, when aerosolized using a third generation e-cigarette, released

ethenone gas, a type of ketene gas and respiratory irritant, which they hypothesized could be

a contributing factor to EVALI (Wu & O’Shea, 2020). Attfield et al. also hypothesized that

thermal degradation of VEA may be important in EVALI and suggested that acetate moieties

are precursors for the formation of ethenone gas and that the reaction occurs at temperatures

in excess of 300 °C in the presence of catalytic metals and/or ceramic surfaces present in

heating coils of e-cigarettes (Attfield et al., 2020). According to Wu & O’Shea, additional

thermal transformation products produced from aerosolized VEA included benzene,

butadiene, and formaldehyde, all of which are respiratory irritants and categorized by

NIOSH as potential occupational carcinogens, and trace amounts of tetrahydrofuran

(respiratory irritant) (NIOSH, 2018); none of these additional transformation products yield

the same pathology observed with EVALI (though whether they have effects as mixtures or

modify other chemical reactions is unknown). Naramani and da Silva used a computational

modeling approach and their results suggested that under “typical” vaping coil temperatures,

VEA was unlikely to produce ketene gases at harmful levels; however, at coil temperatures

above 700 °C, such as might be encountered in a ‘dry hit’, formation of trimethyl quinone

methide acetate would dominate, with lesser amounts of ketene gases produced. The authors

note that even with a low yield of ketene gas, concentrations were predicted to be acutely

toxic in the lungs at coil temperatures above 700 °C (Narimani & da Silva, 2020).

EVALI, whether these other identified aerosolized substances, as mixtures or through

interactions, contributes to the pathophysiology observed for EVALI is unclear.

Another study characterized the chemical composition of bulk liquid diluents (PG, VG,

MCT, squalane, vitamin E, VEA, and triethyl citrate) and their aerosol condensates (Jiang et

chain alkanes (MCT, squalane), and quinones (vitamin E, VEA) in the aerosolized

condensates. Aerosolized vitamin E and VEA yielded the transformation product

duroquinone and aerosolized VEA produced durohydroquinone; both compounds have

capacity to generate ROS. Human lung bronchus epithelial (Beas-2B) cells were exposed

in

vitro

decreases in cell viability compared with their respective bulk liquid. LDH release was

significantly increased for cells exposed to aerosolized condensates of squalane, vitamin E,

and MCT (Jiang et al., 2020). Muthumalage et al. investigated effects of MCT and VEA

IL-17A, IL-12p40, and IL-4. There was no difference in total cells, neutrophil cells, and T-

helper cells in mice exposed to MCT and VEA compared with control. Exposure to VEA

caused changes in eicosanoids, including 6kPGF1a, LTB4, LTC4, LTD4, LTE4, and 5HETE

and a reduction in surfactant protein SP-A levels, the latter which plays an important role in

lipid homeostasis and innate immune defense. Finally, lipidomic profiling revealed that mice

exposed to VEA had significantly higher levels of diradylglycerols, cholesterol ester, and

glycerophosphocholines in BALF compared with controls.

whereas mice exposed to PG/VG contained fewer macrophages and no evidence of lipid

accumulation (Bhat et al., 2020). In another study using C57BL/6 mice (female), Matsumoto

et al. reported that exposure to VEA increased BALF protein levels and plasma surfactant

protein-D levels (which indicated alveolar epithelial injury) compared with air-exposed

controls. VEA exposure also increased total BALF cell counts and neutrophil cell counts

compared with air-exposed controls and mice exposed to aerosolized JUUL® brand flavored

e-liquids; many large vacuolated macrophages contained multiple nuclei (which is a sign of

activation) and stained positive for intracellular lipid droplets. Further, mice exposed to VEA

had significantly increased concentrations of pro-inflammatory neutrophil chemoattractant

products or modifications in consumer behavior and the observed waning trend of new

disease cases (Krishnasamy et al., 2020). Thirdly, using a C57BL/6 mouse model, two

research groups independently reported that inhalation of VEA, but not PG/VG humectants,

caused inflammatory and histological changes in lungs consistent with that observed in

EVALI patients (Bhat et al., 2020; Matsumoto et al., 2020).

VEA is strongly implicated in the EVALI outbreak; however, evidence is not sufficient to

rule out the contribution of other chemicals of concern, including chemicals in either THC

nicotine (Kleinman et al., 2020). E-liquids are chemically complex mixtures so the potential

for unexpected chemistries to occur within an aerosolized mixture is high. Other toxic

substances could have been eliminated from the body prior to BALF collection from EVALI

patients, and MCT and terpenes have capacity to induce oxidative stress and pulmonary

inflammation (Chand et al., 2019; Lozier et al., 2019; Muthumalage, Friedman, et al., 2020;

Wu & O’Shea, 2020).

Type I and Type II epithelial cells, as well as macrophages and polymorphonuclear

leukocyte cells such as neutrophils, eosinophils, and basophils. These innate immune cells

are part of the first line of defense against exogenous exposures and any changes in their

function will alter airway homeostasis (Hickman et al., 2019). Chand et al. proposed the

following sequence of events: 1) aerosolized e-liquid constituents (oils, lipids, VG, and

VEA) are deposited in the alveoli, 2) this chemical insult provokes cell death and the

resulting cellular debris is engulfed by macrophages via efferocytosis leading to

accumulation of lipid-laden macrophages, 3) polymorphonuclear cells are recruited to the

alveoli and neutrophils release extracellular traps (NETs; extracellular fibers to trap

extracellular irritants), and 4) alveolar Type II epithelial and other cells secrete pro-

inflammatory cytokines. Oxidative damage may also contribute to the accumulation of

oxidative derivatives of cellular lipids and lung surfactant in the alveolar region of the lung

which could affect exposure amounts if baseline, unstressed, rates are used in the

calculations; clearance patterns and uniformity of distribution have been demonstrated to be

different depending on the route of exposure whether whole body, nose-only or oral

pharyngeal; animals can becomes stressed because of restraint and no food or water during

the exposure; and, if exposure tubes are used, there is potential for heat and moisture buildup

and animals may try to turn around, which can lead to suffocation (Oberdörster, Castranova,

Asgharian, & Sayre, 2015; Pauluhn & Thiel, 2007; Wong, 2007).

in an experimental system, whether it be a submerged cell culture, ALI system, animal

system, or humans.

9.4. Considerations for biological monitoring

Table 3 summarizes published studies that reported biomarkers of exposure

. Markers

of exposure to aerosolized e-liquids have generally been limited to monitoring a few

substances or metabolites.

S-(3-hydroxy-1-methylpropyl)-L-cysteine (crotonaldehyde metabolite), N-acetyl-S-(3-

hydroxypropyl)-L-cysteine (acrolein metabolite), and N acetyl-S-(3,4-dihydroxybutyl)-L-

cysteine (1,3-butadiene metabolite) were significantly higher in urine of regular tobacco

smokers compared with e-cigarette users (Frigerio et al., 2020). Salivary cytokine

concentrations are being investigated as markers of effect to differentiate between regular

tobacco smokers compared with e-cigarette users for endpoints such as gingival health

(Faridoun, Sultan, Jabra-Rizk, Weikel, Varlotta, and Meiller, 2021). Another possible matrix

for biological monitoring is exhaled breath. Recently, it was reported that e-cigarette users

could be differentiated from regular tobacco cigarette smokers based on presence of esters

(e.g. ethyl acetate), terpenes (e.g. α-pinene, β-pinene, d-limonene, p-cymene, etc.) and

oxygenated compounds (e.g. 3-hexen-1-ol, benzaldehyde, hexanal, decanal, etc.) in their

exhaled breath (Papaefstathiou, Stylianou, Andreou, & Agapiou, 2020). Based on the

paucity of available data, important research gaps/opportunities that relate to biomonitoring

include: 1) what unique markers of exposure for flavoring chemicals that have been shown

to induce toxicity are needed for e-cigarettes? and, 2) what specific markers of effect and

disease need to be identified, validated, and utilized to assess toxicity, and ideally

9.5. Toxicity from passive exposures

In 2010, New Jersey was the first state to prohibit e-cigarette use and vaping in indoor areas

of restaurants, bars, and worksites (Marynak et al., 2017). As of December 2019, 14 states,

the District of Columbia (DC), and Puerto Rico have enacted similar laws (www.cdc.gov/

3

concentrations were 1400–2200 μg/m

3

, and

peak nicotine concentrations were 0.2–0.6 μg/m

3

(Geiss, Bianchi, Barahona, & Barrero-

Moreno, 2015). As inhaled aerosol travels into and out of the body, particles and gases will

deposit in the successive regions of the respiratory tract. Hence, the composition of what is

inhaled as mainstream aerosol differs from what is exhaled as secondhand aerosol (Marco &

Grimalt, 2015; Papaefstathiou, Bezantakos, et al., 2020; Samburova et al., 2018). Several

studies have evaluated concentrations in exhaled breath of e-cigarette users as indicators of

potential for passive exposure. For example, one study reported that the average PG, VG,

and nicotine levels exhaled by an e-cigarette user into a room were 4980, 82,860, and 13.1

μg/m

3

, respectively (Breiev et al., 2016). Another study reported that levels of formaldehyde

exhaled by an e-cigarette user into a room ranged from 5 to 8 μg/m

3

Schripp et al. sampled for 20 VOCs exhaled by e-cigarette users into a room, though only 2-

3

3

3

cigarette users, Saffari et al. quantified 22 metals, 16 alkanes, and 19 organic acids in their

exhaled breath, and Samburova et al. quantified 8 aldehydes in exhaled breath

(Papaefstathiou, Bezantakos, et al., 2020; Saffari et al., 2014; Samburova et al., 2018).

Finally, it was reported that maximum total VOC concentrations were 435–492 ppb when an

e-cigarette user puffed in a room (Tzortzi et al., 2020). These studies demonstrated potential

for secondhand exposure but did not assess whether by standers were actually exposed to

emissions. In the first study of its kind, Flouris et al. directly measured exposure of

volunteers who were passively exposed to e-cigarette emissions exhaled by a user and

reported they had mean serum cotinine level of 2.4 ng/ml, thereby proving that passive

exposure occurs (Flouris et al., 2013). Subsequently, Melstrom et al. measured passive

chemicals in exhaled breath of passively exposed persons and detected seven VOCs (highest

average was for formaldehyde at 14 μg/m

) and nicotine (average of 0.20 μg/m

Drooge, Marco, Perez, & Grimalt, 2019). Tzortzi et al. reported that for passively exposed

persons, changes in VOC concentrations from e-cigarette emissions in a room were

positively associated with reported nasal and throat-respiratory symptoms (Tzortzi,

Teloniatis, et al., 2020). E-cigarette conventions can attract hundreds to thousands of

attendees. Johnson et al. reported that non-smokers who attended several e-cigarette

conventions had elevated urinary cotinine (0.38–1.1 μg/g), salivary cotinine (0.08–0.17

ng/mL) and urinary trans-3′-hydroxycotinine (0.25–0.85 μg/g), S-(3-hydroxypropyl)-N-

acetylcysteine (199–635 μg/g), S-carboxyethyl-N-acetylcysteine (83–117 μg/g), and 8-

Isoprostane (260–466 ng/g) immediately after attending these events (Johnson et al., 2019).

2

cm

ng/100 cm

/h to 69.6 ng/100 cm

nicotine, PG, and VG on surfaces in a room after use of e-cigarettes (Liu et al., 2017). For

more detailed information on passive exposures to e-cigarette emissions, the reader is

referred to a previously published review article (Hess, Lachireddy, & Capon, 2016). The

paucity of laws that prohibit indoor use of e-cigarettes indicate a high potential for

secondhand exposure and additional exposure from residues on surfaces among workers in a

wide range of occupations as well as children and other susceptible populations that reside

or spend time in a space (or possibly an adjacent work space) occupied by someone who

uses an e-cigarette. Much research to date has focused on the characteristics of mainstream

9.5.1. Research gap/opportunity 11

transient ocular, nasal, throat-respiratory irritation symptoms (Tzortzi, Teloniatis, et al.,

2020). Mathematical modeling of secondhand exposures have included residential and

occupational exposure scenarios to aldehydes and ultrafine particles (Avino et al., 2018;

Colard, O’Connell, Verron, Cahours, & Pritchard, 2014; Logue et al., 2017; Rostami,

Agyemang, & Pithawalla, 2018; Rostami et al., 2016). The importance of understanding

toxicity in susceptible populations from passive exposures was raised by possible adverse

health effects such as bronchodilation of neonatal bronchial rings (see Section 7.3).

Additionally, adults exposed to secondhand aerosol experienced respiratory irritation and

reviewed literature documented activation of respiratory irritation signals (see Section 7.5).

represents mainstream aerosol. Hence, a critical research gap/opportunity is how to generate

a realistic secondhand aerosol for toxicity testing of flavored e-liquids. A possible solution

may be to reproducibly generate mainstream aerosol using a smoking machine and pass it

through a tube coated with surfactant to mimic aerosol interactions with lung airway lining

fluid (Davies et al., 2017; Sosnowski et al., 2018) to obtain a surrogate secondhand aerosol.

In the future, it may be possible to use 3-dimensional printed lung models to better mimic

aerosol interactions with the respiratory tract compared with a simple tube model.

9.5.2. Research gap/opportunity 12

Does exposure to exhaled e-cigarette aerosols that form residues on surfaces result in

toxicity to receptors?: Studies of exposure potential have documented accumulation of

9.6. Toxicology of EVALI

9

appropriate test system, 2) creation of surrogate e-liquids with reasonable similarity to those

that elicited a toxic response in cases, 3) characterization of a meaningful vaping topography

that is representative of cases, 4) generation and characterization of gas and liquid phase

constituents for representative puff topographies, and 5) development of representative dose

estimates to recreate conditions that yielded EVALI cases. Toward this last gap, De Jesús et

al. recently published analytical methods to determine terpenes and petroleum distillates in

BALF, which will enable improved dose modeling for EVALI (De Jesús et al., 2020; De

Jesús, Silva, Newman, & Blount, 2020).

properties of aerosolized e-liquids in EVALI cases are needed to understand regional lung

deposition of liquid droplets. Characterization of chemical thermal degradation products in

aerosolized e-liquids from EVALI cases is important to understand the formation of

secondary reaction products that may be harmful or react to form harmful compounds.

Additionally, the partitioning of hazardous substances between the gas and liquid droplet

phases is important for understanding dose. Mechanistic toxicology studies are needed to

better understand this disease process, identify biomarkers of effect, and reduce future

morbidity and, in some cases, mortality.

inhibits many powerful components of the body’s respiratory defense

mechanisms against inhaled foreign materials.

Cardiovascular and circulatory systems – Some aerosolized flavored e-liquids

Developmental effects – Specific flavored e-liquids were more cytotoxic to

human embryonic stem cells compared with differentiated adult HPF cells;

among flavorings, cinnamaldehyde and 2-methoxycinnamaldehyde were the

most cytotoxic flavoring constituents. Additionally, some flavored e-liquids

induced bronchodilation of neonatal but not adult bronchial rings, which

Skeletal system – Some aerosolized flavored e-liquids increased biofilm

formation and decreased enamel hardness of teeth compared with aerosolized

unflavored e-liquids. Aerosolized flavored e-liquids that contained ethyl butyrate,

hexyl acetate, and triacetin were associated with bacteria-initiated

demineralization of enamel and ethyl maltol inhibited

tongue squamous cells to certain e-liquids indicated potential risk for

development of HNSCC. Menthol flavoring in e-liquids was shown to induce

cellular DNA damage and chromosomal damage

volunteers who used a tobacco-flavored e-liquid without nicotine had

significantly higher levels of micronucleus formation in oral buccal cells

compared with air-exposed controls. In summary, exposure to certain flavored e-

liquids can induce genotoxic or mutagenic responses in several cell models and

in humans, which raises the possibility for cancers.

EVALI – VEA is strongly linked to EVALI; however, the mechanism or

mechanisms by which VEA causes EVALI is currently unclear. One hypothesis

Acknowledgements

The authors would like to thank P. Erdely and D. Weissman at NIOSH and J. Fowles at the California Department

of Public Health for critical review of this manuscript prior to its submission to the journal. The authors also thank

JR Wells and ED Arnold at NIOSH for helpful discussions on carbonyl and flavorings chemistry.

Funding Source

Financial support for this review was from NIOSH intramural funds. The funding source was not involved in the

conduct or preparation of the review or in the decision to submit the paper for publication.

List of abbreviations

-THC

2