Harmful and Potentially Harmful Constituents in E-Liquids and Aerosols from Electronic Nicotine Delivery Systems (ENDS)

Abstract

In 2012, the U.S. Food & Drug Administration (FDA) published an established list of 93 harmful and potentially harmful constituents (HPHCs) targeting four tobacco product types (cigarettes, cigarette tobacco, roll-your-own tobacco, smokeless tobacco). In 2016, the FDA finalized the deeming rule to regulate electronic nicotine delivery systems (ENDS). However, knowledge gaps exist regarding whether certain HPHCs are present in ENDS e-liquids and aerosols. We identified and addressed these gaps by conducting literature searches and then experimentally quantifying HPHCs in the e-liquid and aerosol of 37 ENDS brands based on gaps in the literature. The literature searches identified 66 e-liquid HPHCs and 68 aerosol HPHCs that have limited to no information regarding the quantifiability of these constituents. A contracted ISO 17025 accredited laboratory performed the HPHC quantifications. The availability of validated analytical methods in the contracted laboratory determined the HPHCs included in the study scope (63/66 for e-liquids, 64/68 for aerosols). Combining the results from the quantifications and literature searches, 36 (39%) and 34 (37%) HPHCs were found quantifiable (≥limit of quantification [LOQ]) in ENDS e-liquids and aerosols, respectively, with 25 HPHCs being quantifiable in both matrices. Quantifiability results imply potential HPHC transfers between matrices, leaching from components, or formations from aerosol generation. The study results can inform the scientific basis for manufacturers and regulators regarding regulatory requirements for HPHC reporting. The HPHC quantities can also inform evaluations of the public health impact of ENDS and public communications regarding ENDS health risks.

Introduction

In 2016, under the authority of the U.S. Federal Food, Drug, and Cosmetic Act (FD&C Act), the U.S. Food and Drug Administration (FDA) issued a final rule to deem products meeting the statutory definition of “tobacco product” to be subject to the FD&C Act.¹ This rule extends the FDA’s authority to regulate the manufacture, marketing, and distribution of deemed tobacco products, which include electronic nicotine delivery systems (ENDS), to protect the public health.¹ The FD&C Act also requires tobacco product manufacturers and importers to report harmful and potentially harmful constituent (HPHC) quantities in ENDS, including yields in aerosol. HPHCs are chemicals (including chemical compounds) in a tobacco product or its aerosol that (a) are or potentially are inhaled, ingested, or absorbed into the body and (b) cause or have the potential to cause harm to tobacco product users or nonusers.² FDA published an established list containing 93 HPHCs in 2012³ and, in 2019, proposed adding 19 additional HPHCs relating specifically to ENDS and other deemed tobacco products.⁴

A recently published study reported the quantities of the 93 HPHCs in tobacco-containing tobacco products (TCTPs, i.e., cigarettes, roll-your-own [RYOs], smokeless tobacco products [STPs], cigars, pipes, waterpipes).⁵ However, knowledge gaps remain regarding which HPHCs are reliably quantifiable (i.e., at and above the limit of quantification [LOQ] of the analytical test method) in ENDS. ENDS differ from TCTPs in ingredients, composition, processing, manufacturing, and usage. Typically, ENDS contain e-liquid rather than tobacco. E-liquids contain carrier solvents (e.g., propylene glycol [PG], vegetable glycerin [VG]), nicotine (i.e., tobacco-derived, nontobacco-derived, synthetic), and flavors. To deliver nicotine, ENDS devices heat e-liquids into aerosols that users inhale, rather than burning tobacco to form smoke. ENDS e-liquid and aerosol chemistry profiles are also complicated by their variable e-liquid formulations including flavors (e.g., tobacco, menthol, fruit),⁶ device designs (e.g., disposables, cartridges, pods, refillable devices),⁷ nicotine forms (e.g., free, salt),⁸⁻¹⁰ and product characteristics and uses (e.g., power setting,^11,12 coil temperature,¹² puffing behaviors^7,13,14). Thus, the quantifiability of HPHCs in ENDS is likely different from that in TCTPs. Furthermore, little information is available regarding how ENDS HPHC quantities vary across marketed ENDS product by brands and by matrix (i.e., e-liquid, aerosol).

When examining HPHCs in ENDS, both the exposure and the exposure route (i.e., aerosol inhalation) to users and nonusers should be considered. An HPHC may be quantifiable in the finished ENDS product (i.e., in e-liquid or components and parts), but if it does not reach users and nonusers, the potential harm associated with this HPHC in the ENDS may be minimal. Aerosol HPHCs transferred from e-liquids or product components and parts or generated during the aerosol formation process may warrant greater attention. One example is the formation of aerosol carbonyls from the heating¹⁵ of carrier solvents¹⁶ and flavors^7,17,18 in e-liquids. Thus, it is important to examine both ENDS e-liquids and aerosols.

This study analyzes various ENDS brands and designs while keeping certain product performance settings (e.g., puffing regimen, wattage) consistent to determine which HPHCs are quantifiable in ENDS e-liquids and aerosols and to compare these results to TCTPs. Specifically, this study (1) identifies the knowledge gaps regarding the quantifiability of HPHCs by examining available literature to identify HPHCs (with limited or no data available; see the Methods section for details) in ENDS e-liquids and aerosols and (2) addresses these knowledge gaps by quantitatively measuring the HPHCs in ENDS e-liquids and aerosols. This study reveals which HPHCs are quantifiable and which are not (<LOQ of the analytical test method) in ENDS e-liquids and aerosols and compares the findings to previous findings for TCTPs.⁵

Methods

HPHC Groupings

Similar to our previous work,⁵ this study splits the 93 HPHCs on the 2012 established list³ into 12 groups: alkaloids, amides, aromatic amines, carbonyls, chlorinated dioxins/furans, heterocyclic aromatic amines, metals, nitrosamines, phenols, polycyclic aromatic hydrocarbons (PAHs), volatiles and semivolatiles, and others. Each group consists of compounds that have similar chemical properties and are typically quantified using similar analytical methods (though methods used may differ between laboratories and matrices), except for the “others” group, which encompasses all HPHCs not fitting into any other groups. Among the 93 HPHCs, cresols and chlorinated dioxins/furans are compound classes. This study examines o-cresol, m-cresol, and p-cresol to represent cresols, and 17 specific and 8 groups of dioxins/furans (listed in Tables S5 and S6) to represent chlorinated dioxins/furans based on similar testing performed in the pharmaceutical and environmental industries. Thus, the study scope includes 119 individual constituents.

Knowledge Gap Identification

We conducted two literature searches and reviews to identify knowledge gaps regarding the quantifiability of the 93 HPHCs in ENDS e-liquids and aerosols. The first search and review were conducted between 2014 and 2018 for publications up to 2018. The second search and review were conducted in 2020 to capture any additional literature published between 2018 and 2020. The databases for both literature searches included SciFinder, PubMed, EBSCOHost, Web of Science, and Embase. Keywords included “constituent” OR “chemical (specific HPHC name)” OR “chemistry” AND “electronic cigarette” OR “ENDS” OR “e-liquid.” The two literature searches were not exhaustive, but rather constituted broad examinations of the literature. After excluding patents, presentations, articles not in English, and articles not relating to HPHCs in ENDS, 56 articles (i.e., 15 and 41 articles from each search; see Table S1) were reviewed and found to contain gap identification information. The first search results place each HPHC in each matrix into one of three categories: “limited data,” “no data,” and “established data.” An HPHC was categorized as having “limited data” if (1) the publication(s) reviewed did not study domestically (i.e., United States) available commercial ENDS, (2) the publication(s) lacked test method information, or (3) the published methods were more vulnerable to artifacts and false findings (e.g., lower sensitivity and specificity). An HPHC was categorized as having “no data” if the literature search resulted in zero publications for the HPHC for the matrix of interest. Lastly, an HPHC was categorized as having “established data” if the HPHC was not in the “limited data” and “no data” categories. For “established data,” if many publications were available for a specific HPHC, the categorization gave more weight to publications that tested multiple brands using accredited laboratories and methods. “Limited data” and “no data” are considered to be knowledge gaps. The second search results further supported the findings (i.e., gaps) from the first search. Table S1 shows the categorization results with references from both searches. This study quantitatively measured the ENDS HPHCs in the “limited data” and “no data” categories.

Analytical Test Methods

After knowledge gaps were identified from the literature searches, validated analytical method availability in the testing laboratory, which is competent by International Organization for Standardization (ISO) 17025 and has extensive experience in HPHC testing, and its subcontractor determined which HPHCs to address. Fully validated analytical methods or analytical methods within the scope of the laboratory quality system were considered available test methods for the study. Enthalpy Analytical, an ISO 17025 accredited laboratory, was contracted to perform the quantification of the HPHCs between 2019 and 2023 across its three laboratory locations and one subcontractor (GEL Laboratories for radioactive metals). All quantitative methods were proprietary and fully validated for the intended purpose (i.e., contained all needed validation parameters: accuracy, precision, LOD, LOQ, specificity, range, linearity, robustness). In addition, the methods were either in the laboratory’s accreditation scope, under a quality control system, or subcontracted to a specialized laboratory. As a general practice, the contract laboratory adopts CORESTA Recommended Methods (CRMs) and ISO standards and uses modern testing technologies. Table S3 lists the instrumentation used for each HPHC in each matrix. All methods presented in Table S3 are quantitative and fully validated unless otherwise noted. LODs and LOQs are listed in Tables S5 and S6.

For each HPHC, Enthalpy Analytical conducted five replicate measurements per study design for each matrix of each product brand. In rare cases of logistical difficulties, if the allocated product quantity did not allow for at least three replicates, the specific HPHC was not measured for that product brand (Tables S4–S8 note when this occurred). However, for the radioactive metal analysis subcontracted to GEL Laboratories, only one replicate was performed per brand (noted in Tables S4–S8) due to the limitations GEL Laboratories had at the time. For all aerosol analyses, the puffing regimen used a 55 mL “puff volume,” a 3-s “puff duration” and a 27-s puff interval that sums to a 30-s “puff period” with a rectangular shape “puff profile” following ISO 20768 specifications.¹⁹ The ENDS devices were set at 100% open airflow, where applicable. Aerosol collections started at puff 16 after 15 priming puffs with all devices positioned at a 45° angle, batteries fully charged, e-liquids refilled/replaced, wattages set at 30 W (for open devices), and new coils, where applicable. The conditioning and aerosol collection conditions (i.e., temperature, relative humidity) were within the ranges specified in the testing laboratory standard operating procedure (SOP) and met requirements and specifications as defined in ISO 20768. End-of-life (EOL) studies were conducted for all ENDS brands to determine the appropriate puff counts per “puff block” (i.e., “finite series of sequential puffs”; for example, puffs 1 to 50 and puffs 51 to 100 are two different puff blocks)²⁰ for aerosol collections (Table S4).

EOL and coil temperature (CT) analyses used the same puffing regimen and profile as described above for aerosol collections. The EOL analysis machine puffed the ENDS until the assumed EOL (i.e., approximately 85% e-liquid depletion, 150 or 210 puffs) and collected aerosol collected mass (ACM) on a Cambridge filter pad (CFP) to determine the specific puff count(s) and puff block(s)²⁰ that generated the highest ACM for the HPHC analyses. The maximum number of puffs (rounded down to the nearest interval of five) that could be collected on each CFP was calculated to keep the ACM below the CFP loading capacity (600 mg). The “maximum puff block/count” was defined to be the puff block/count that generated the highest ACM. The “maximum ACM puff sessions” were determined based on the ACM generated from 15 puffs that constitute a puff session, starting at puff 16, 106, and 196 for products with EOL of 210 puffs and at puff 16, 76, and 136 for products with EOL of 150 (see Table S4). The CT data were collected from only one replicate for each brand of one battery and device unit during activation using a temperature data logger with a thermocouple attached directly to the ENDS coil. The CT analyses were performed independent of HPHC generation. Depending on the ENDS product, 25–60 puffs were evaluated for each product with approximately 220 temperature readings collected per puff. The mean temperature is the average of all data points collected for each product while the maximum and minimum temperatures are the highest and lowest temperatures among the means of the approximately 220 data points from each puff unless otherwise noted in Table S6. The CT analyses mirror the HPHC collection conditions.

Product Brands and Sample Matrix Selection

This study selected 37 ENDS products that reflected both high and low market shares to represent the U.S. ENDS marketplace based on ENDS product data graphics retrieved from Euromonitor International Passport in July 2018. The marketplace shares generally reflect manufacturer sizes (i.e., large and small manufacturers). The selection also considered products receiving public attention (e.g., appearing in news articles or online store discussion forums) or having certain characteristics (e.g., flavors, nicotine forms, nicotine concentrations, product performance designs). For example, the study aimed to include a minimum of two ENDS flavors (i.e., tobacco, menthol) for each brand and a maximum of two additional flavors (e.g., fruit, sweet), if logistically feasible. Products with different designs (e.g., disposable, refillable and rechargeable, variable power, cartridge, pod, pod-mod, sub-ohm mods) were included and determined from product descriptions. For device-only products, e-liquids were selected based either on device manufacturer recommendations or suitability of the e-liquid formulation. For example, e-liquids containing free nicotine at relatively lower concentrations were selected for sub-ohm mods devices based on manufacturer recommendations. Table S4 shows the studied products and related product information. The brands we studied differed slightly from the brands listed in the original study plan due to changes in marketplace availability. To minimize the impact of the brand changes, we selected the replacement brand that matched the original brand in flavor, product design, nicotine strength, nicotine form, manufacturer, or/and market size, when possible.

Bizzell Group LLC (Bowie, Maryland) acquired the ENDS brands in 2019–2020. This study required that all products be from the U.S. market but did not specify a sampling procedure for either product acquisition (e.g., specific geographical locations within the U.S.) or sampling at the time of testing. The actual product acquisition procedure at the product purchasing stage was similar to that of “at point of sale at one time” specified in ISO 8243.²¹ The sampling procedure at the time of testing followed the established sampling protocols of the testing laboratory. Samples were stored and maintained at ambient temperature until testing unless the protocol required other storage beforehand (e.g., chlorinated dioxins/furans e-liquid testing required storage at −20 °C). All samples were visually inspected before testing to ensure that no change was observed in e-liquids during storage. Two JUUL products (i.e., Fruit, Crème) were taken off the U.S. market during our product procurement; as a result, the acquired quantity was sufficient for testing in e-liquids only. To mitigate the impact of this occurrence, we used two JUUL products (i.e., CA Fruit, CA Vanilla) from the Canadian market for aerosol HPHC testing. As it is unknown whether the products of the same flavor from the U.S. and Canadian markets were identical, data from these products were not used to evaluate the e-liquid-to-aerosol transfer.

Data Quality, Processing, and Statistics

Data generation quality control occurred at the testing level via the ISO 17025 accredited laboratory quality control system. Data quality checks were performed to ensure that all data were accurate, reliable, error-free, and processed consistently. These data quality checks included one researcher examining the contractor-reported results to compare the raw data to the summarized data in order to (1) ensure no transcription errors, (2) ensure that the mean and standard deviation were correctly calculated, (3) determine that the values labeled as below the LOD (BDL) or below the LOQ (BLOQ) were correctly labeled, and (4) ensure that the tobacco product name and description were correctly labeled. The researcher also examined the test information, including method validation parameters, to ensure that the information supported the claimed method validation status and reported data. Tables S5 and S6 present the data summaries for e-liquids and aerosols, respectively. After the data quality check, the mean HPHC quantity or yield for each brand set was categorized as quantifiable or not quantifiable. For this study, excluding radioactive metals, “quantifiable” is defined as at least three (out of five) replicates of a brand having a quantity or yield at or above LOQ, while “not quantifiable” is defined as not meeting this criterion. For radioactive metals, quantifiable and not quantifiable categorization is defined directly from the single replicate value reported by the subcontracted laboratory. Once quantifiable and not quantifiable categories were coded, the percent quantifiable for each HPHC for each matrix was calculated based on the number of brands assessed. For example, anabasine in ENDS e-liquid was quantifiable in 15 of the 37 brands examined; thus, the percent quantifiable is 41%. All data quality checks and the HPHC categorizations were independently verified. We used JMP 15 to create heatmaps depicting the calculated quantifiable percentage for each quantifiable HPHC by brand in each matrix (Figure 1).²²

Figure 1.

Heatmap of HPHCs in the e-liquids and aerosols studied. The color scale matches the percentages of brands with quantifiable quantities and yields summarized in Tables S5 and S6 (yellow = 0%; navy = 90%).

Results

Knowledge Gap Identification

The literature review revealed that, despite the exponential growth of ENDS research between 2014 and 2019, there were either no or limited published data regarding the quantifiability of 66 out of the 93 HPHCs in ENDS e-liquids (Table S1). Of these 66 HPHCs, the contracted laboratory had validated test methods to analyze 63 (Tables S1 and S3). Similarly, there were either no or limited published data regarding the quantifiability of 68 out of 93 HPHCs in ENDS aerosols (Table S1), of which the contracted laboratory had validated test methods to analyze 64 (Tables S1 and S3). However, this study reports the results for 67 HPHCs in ENDS aerosols, as the contracted laboratory also measured 1,3-butadiene, acrylonitrile, and isoprene, which have “established data,” concurrently with other in-scope volatile organic compounds (Table S3).

For HPHCs considered having established data following the literature search, an HPHC was considered quantifiable if at least one publication reported the HPHC at or above LOQ in the ENDS matrix of interest. For example, benzo[a]pyrene (BAP) in ENDS aerosol is considered quantifiable in ENDS aerosol, although only one²³ of the six investigated publications²³⁻²⁸ reported BAP at quantifiable yields in ENDS aerosol. The decision to be conservatively inclusive was made because each publication could examine only a small fraction of the vast ENDS market, which contains wide differences in device designs and e-liquid formulations including ingredients, materials, and performances. Of the 27 e-liquid HPHCs and 25 aerosol HPHCs having established data, the majority of publications agreed on HPHC quantifiability for all e-liquid HPHCs and 21 aerosol HPHCs. However, a majority of publications did not agree on quantifiability for four aerosol HPHCs (1-aminonaphthalene [1/4], 2-aminonaphthalene [1/4], 4-aminobiphenyl [1/4], and BAP [1/6]). Table 1 presents the publications that support the HPHC quantifiability categorizations, while Table S1 presents all references from the literature search for quantifiability determination.

Table 1. HPHCs Quantifiable and Not Quantifiable in ENDS E-Liquids and Aerosols Based on the Results of This Study and the Literature^a.

	e-liquid	aerosol	both matrices
Quantifiable HPHCs	- Methyl Ethyl Ketone (MEK)	- Acetamide	-Anabasine	- Nickel(6,30,34,35,37,40,50−53,55)
	- Arsenic	- 1-Aminonaphthalene(31)	- Nicotine(34,35)	- 4-(Methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK)(29,34−37,40,58−67)
	- Selenium	- 2-Aminonaphthalene(31)	- Nornicotine	- N-Nitrosonornicotine (NNN)(34−37,40,57,58,60−68)
	- Benz[a]anthracene (BAA)(29)	- 4-Aminobiphenyl(31)	- Acrylamide	- Catechol
	- Benzo[b]fluoranthene (BBF)(29)	- Mercury	- o-Toluidine	-m-, o-, p-Cresol(29)
	- Benzo[k]fluoranthene (BKF)(29)	- N-Nitrosodimethylamine (NDMA)	- Acetaldehyde(11,14,25,34−48)	-Phenol(29)
	- Chrysene(29)	- Propylene Oxide	- Acetone(14,29,34,41,42,45−48)	- Benzo[a]pyrene (BAP)(23,29)
	- Dibenz[a,h]anthracene (DahA)(29)	- Caffeic Acid	- Acrolein(11,25,34−40,43,44,46,48)	- Benzene(29,30,35)
	- Indeno[1,2,3-cd]pyrene (IcdP)(29)	- Carbon Monoxide (CO)(7,32,33)	- Crotonaldehyde(35,38,42,47,48)	-Styrene(29,30)
	- Naphthalene(29)		- Formaldehyde(11,14,34−48)	- Toluene(12,29−31,35,57,69)
	- Ethylbenzene(29,30)		- Propionaldehyde(25,29,44−46,48,49)
			- Cadmium(6,26,30,34,35,40,50−53)
			- Chromium(6,26,30,31,35,37,51−57)
			- Cobalt
			- Lead(6,26,34,35,37,40,51−53)
Not Quantifiable HPHCs	- Acetamide	- 2,6-dimethylaniline(57,67)	- 2-Amino-9H-pyrido[2,3-b]indole (AαC)	- 5-Methylchrysene
	- 1-Aminonaphthalene	- MEK	- 2-Amino-6-methyldipyrido[1,2-a:3′,2′-d]imidazole (Glu-P-1)	- Benz[j]aceanthrylene (BJA)
	- 2-Aminonaphthalene	- Arsenic	- 2-Aminodipyrido[1,2-a:3′,2′-d]imidazole (Glu-P-2)	- Benzo[c]phenanthrene (BCP)
	- 4-Aminobiphenyl	- Selenium	- 2-Amino-3-methylimidazo[4,5-f]quinoline (IQ)	- Cyclopenta[c,d]pyrene (CcdP)
	- Mercury	- BAA	- 2-Amino-3-methyl-9H-pyrido[2,3-b]indole (MeAαC)	- Dibenzo[a,e]pyrene (DaeP)
	- NDMA	- BBF	- 2-Amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)	- Dibenzo[a,h]pyrene (DahP)
	- N-Nitrososarcosine (NSAR)	- BKF	- 3-Amino-1,4-dimethyl-5H-pyrido[4,3-b]indole (Trp-P-1)	- Dibenzo[a,i]pyrene (DaiP)
	- Propylene Oxide	- Chrysene	- 1-Methyl-3-amino-5H-pyrido[4,3-b]indole (Trp-P-2)	- Dibenzo[a,l]pyrene (DalP)
	- Aflatoxin B1	- DahA	- Beryllium	- Polychlorinated Dibenzo-p-Dioxans (PCDDs)
	- Ammonia	- IcdP	- Polonium-210	- Polychlorinated Dibenzofurans (PCDFs)
	- Caffeic Acid	- Naphthalene	- Uranium-235	-1,3-Butadiene(24,25,27,28,37,57,62)
	- Coumarin	- Ethylbenzene	- Uranium-238	- 2,3-Benzofuran
		- Ammonia(24,27,31,57)	- N-Nitrosodiethylamine (NDEA)	- 2-Nitropropane
			- N-Nitrosodiethanolamine (NDELA)	-Acrylonitrile(24,25,27,28,35,57,62)
			- N-Nitrosomethylethylamine (NMEA)	- Ethylene oxide
			- N-Nitrosomorpholine (NMOR)	- Furan
			- N-Nitrosopiperidine (NPIP)	-Isoprene(24,25,27,28,57)
			- N-Nitrosopyrrolidine (NPYR)	- Nitrobenzene
				- Nitromethane
				- Quinoline
				- Vinyl acetate
				- Vinyl chloride
				- Ethyl Carbamate
				- Hydrazine
				- Hydrogen Cyanide (HCN)
HPHC Presence Unknown	- 2,6-dimethylaniline	- NSAR	- o-Anisidine
	- CO	- Aflatoxin B1
		- Coumarin

^aBold Typeface: Measured in this study only. Italic Typeface: Presence and absence determined based on the literature search results only. Bold Italic Typeface: Measured in this study and based on the literature search results. Default Typeface: Not measured in this study and has been determined to have “Limited Data” to “No Data” from literature search results.

EOL Determination and CT Measurement

The maximum puff count and the maximum puff block for the HPHC analysis were determined for each brand prior to HPHC measurement. The maximum puff count ranged from 20 to 90 puffs, varying by brand and sub-brand (see Table S4). The maximum puff blocks could occur at either the “beginning,” “middle,” or “end” of the product use lifecycle (see Table S4). When a beginning block was not the highest ACM generation puff block but did not significantly differ in ACM from the highest puff block according to a simple statistical two-tailed t test at 95% confidence level, the beginning puff block was selected for aerosol collection. This approach applied to 14 of the 37 products (see bold text in Table S4). However, HPHCs for two JUUL sub-brands were collected at the beginning block, although their end block was the maximum puff block, which statistically differed in ACM from the beginning block (see bold text in Table S4). The contractor did not justify this deviation from the study design. Thus, for these two JUUL products, the HPHC values observed may be underreported because of the relatively smaller amounts of aerosols collected for analysis. Overall, the HPHCs of most brands were collected during the beginning (n = 30) block while the HPHCs of only a small subset of brands were collected during the middle (n = 5) or end (n = 2) blocks.

Working coil temperature may provide additional insight into HPHC generation and quantities, although limited, as the working coil temperature was measured across multiple puffs from one single device and battery for each product (see the Methods and Discussion sections for more details). Depending on the brand, 25–60 puffs were evaluated with approximately 220 data points collected per puff. The mean temperatures were 48.7–150 °C across all brands with a temperature range of 25.6–217 °C (see Table S6). These results provide an estimated range of temperatures each brand could potentially reach, as they are based on single replicate measurements. The measurements do not reflect the temperature variation originating from the within-product manufacturing variations, as noted in other publications.^15,70,71 However, within-product variation may be discerned by comparing sub-brands (with different characterizing flavors) from the same brand. For example, for the U.S. JUUL brands, the mean temperatures ranged from 73.3 to 150 °C. Five products had a minimum temperature close to ambient temperature, which was observed during first time points collected. This potentially indicates a slow heat-up rate for the product. As all five products are either disposable or replaceable cartridge systems, the slow heat-up rate may be related to product designs.

HPHC Testing, ACM, and e-Liquid Weight Loss (EWL)

Tables S5 and S6 summarize the measurement results for HPHCs in e-liquids and aerosols, respectively. The tables include the measured mean and standard deviation for each HPHC in each brand as well as the mean, standard deviation, and range for each HPHC across 37 brands. Results reveal that nine of the 63 HPHCs analyzed in e-liquids and 14 of the 67 HPHCs analyzed in aerosols (64 HPHCs within the study scope + three additionally reported HPHCs) were at or above LOQ. Figure 1 is a heatmap depicting the percentage of brands for HPHCs examined in this study that were quantifiable in e-liquids and aerosols. When combining these results with the literature search data, in total, 36 and 34 of the 93 HPHCs were at or above LOQ for ENDS e-liquids and ENDS aerosols, respectively (see Tables 1 and S2). Table 1 summarizes the HPHCs that were quantifiable, not quantifiable, or with remaining information gaps in e-liquids and aerosols. Table 1 also notes whether the finding was based on literature, results of this study, or both.

Tables S7 and S8 summarize the measured ACMs and EWLs, respectively, from HPHC and EOL analyses. Mean ACM strongly and linearly correlates with the mean EWL for each brand across all HPHC experiments (r = 0.999, Figure 2A). The coefficients of variation (CVs) also matched between ACM and EWL by brand (r = 0.974, Figure 2B), excluding two outliers: the mercury analysis of Vuse Alto Rich Tobacco and the radioactive metal analysis of MyBlu HoneyMoon. The large EWL CVs (44 and 32%) for these two products were driven by the high EWL (14.6 ± 3.3 mg/puff, Z-score 3.93) and low EWL (0.15 ± 0.243 mg/puff, Z-score −3.04) compared to other EWLs (4.80 ± 0.46 and 6.61 ± 1.35 mg/puff) for the same product. By removing these outliers, the EWL CVs are 10 and 20% and have excellent agreement with ACM CVs (9 and 21%). These Pearson correlations are also shown in Figure 2.

Figure 2.

Pearson correlations (r) of aerosol collected mass (ACM) and e-liquid weight loss (EWL) across all HPHC experiments comparing both (A) the overall means and (B) the coefficients of variation (CVs). EWL was selected as the x-axis because it is the source of the generated aerosol. Each data point represents an average of all (A) means and (B) CVs observed for each product across 16 different HPHC measurements, respectively. EWLs from mercury analysis for Vuse Alto and radioactive metals analysis for MyBlu HoneyMoon were removed from (B) as both data points are statistical outliers (z-scores = 3.93, −3.04).

Discussion

ENDS Parameter Results

ENDS brands vary in aerosol delivery by product design. As demonstrated by the EOL results (Table S4), the maximum number of puffs collected without overloading the CFPs ranges from 20 to 90 puffs across 37 brands. This maximum puff number correlates with ENDS designs at certain levels. Specifically, ENDS advertised as disposables, replaceable cartridges, and replaceable pods need more puffs (60–90 puffs) than ENDS advertised as refillable tanks and sub-ohm mods (20–60 puffs) to reach the CFP loading capacity, while ENDS advertised as pod-mods have higher variability in collected puffs needed (25–90 puffs). The maximum puff block occurring in the product lifecycle also varies by ENDS design. The maximum puff block was always at the beginning of the product use lifecycle for four examined disposable ENDS and four examined replaceable cartridge ENDS. One replaceable cartridge ENDS generated similar aerosol amounts (i.e., not statistically different) between the beginning and middle of the product use lifecycle. However, the maximum puff block tends to be at the second half of the product use lifecycle (middle, end) for the 10 examined replaceable prefilled pod or tank ENDS. The maximum puff block occurred variably during the product lifecycle for the 19 examined sub-ohm mods and pod-mods ENDS. The EOL studies are limited as assumptions were made regarding the total puff numbers of each device (Table S4). Thus, aerosol delivery information from puffs after 150 or 210 may not have been captured. For aerosol HPHC analysis, the selected puff blocks were either the maximum puff blocks or the puff blocks that generated similar (not statistically different) amounts of aerosol compared to the maximum puff blocks. Two exceptions are the ENDS where the selected puff block was not the maximum puff block (see the Results section). Additionally, the selection of the aerosol collection puff block assumes that HPHC quantity is positively correlated with ACM. However, some constituents may not be correlated with ACM or should be measured through the whole product lifecycle for data reliability.^37,72 Despite these limitations, the EOL study findings stress that device type needs to be considered when assessing ENDS aerosols.

When comparing the ACM from the selected 15-puff session from the EOL study to the overall EWL and ACM from the aerosol HPHC analyses, the per-puff masses correlated well (r = 0.956 and 0.955, respectively), suggesting that such an EOL study can be a useful tool when selecting the intended ACM and EWL ranges for HPHC analyses. However, EOL results do not replace ACM or EWL measurements for specific HPHC analyses, as an EOL study cannot capture potential issues (e.g., leaking) that may occur in experimental replicates. The overall ACM and overall EWL are strongly and linearly correlated (r = 0.999), indicating process consistency in aerosol generation across 16 of the 17 HPHC analyses (Tables S7–S8; metal analyses are not included, as ACM could not be collected due to the nature of the analyses). Thus, ACM and EWL agreement can be an indicator of a reliable aerosol generation process. The strong linear correlation between the EWL and ACM CVs (r = 0.974) also indicates a well-controlled aerosol generation process. However, higher but matched EWL and ACM CVs may indicate lower product performance consistency in aerosol delivery.

The CT results (Table S6) show the wide average temperature range (25.6–217 °C) at which the examined 37 brands operated, with the maximum temperature trending at the lower end of the range for disposable or replaceable cartridge brands (64.2–133 °C), at the middle of the range for replaceable prefilled pods (87.5–192 °C), and at the high end of the range for pod-mod and sub-ohm brands (103–217 °C). The maximum operating temperature for each brand was lower than 300 °C, suggesting a low likelihood of dry puffing during HPHC testing.^73,74 The averages, standard deviations, and ranges of the coil operating temperature suggest that the examined products vary across puffs in a brand’s ability to consistently reach elevated temperatures, in heat dissipation, and the proximity of coil temperature to room temperature after a puff. ENDS that dissipate heat well and consistently reach elevated temperatures (e.g., Mistic Black Menthol) have an average coil temperature close to the average of the minimum and maximum values observed. Conversely, ENDS that do not dissipate heat well (e.g., JUUL Virginia Tobacco) have an average closer to the maximum, while those that do not consistently reach elevated temperatures (e.g., JUUL Crème) have an average closer to the minimum. While this experiment provides some insight into the brands’ coil temperatures, the results are preliminary and limited. The experiment is a point (location) measurement using thermocouples with probes inserted into ENDS devices and attached to heating coils, which may interfere with normal operation.^71,73,75 Furthermore, as replicability was limited to multiple puffs within the same device, the results may reflect only the device being examined and not the corresponding product itself (e.g., JUUL Crème), if there is variation across devices within the product line.^15,71

ENDS E-Liquid HPHCs

Among the 63 HPHCs analyzed, nine HPHCs (acrylamide, anabasine, arsenic, catechol, cobalt, methyl ethyl ketone [MEK], nornicotine, o-toluidine, selenium) are quantifiable in 3–41% of the 37 e-liquid brands examined (Table 1, Table S5). Three of the nine HPHCs have much higher averages compared to method LOQs: anabasine (average: 26.6 ± 17.0 μg/g; LOQ: 8.79 μg/g), nornicotine (average: 15.0 ± 5.1 μg/g; LOQ: 8.53 μg/g), and o-toluidine (average: 1004 ± 964 pg/g; LOQ: 286 pg/g). Anabasine and nornicotine are nicotine-related impurities (i.e., minor tobacco alkaloid⁷⁶ and nicotine microbial degradation product,⁷⁷ respectively), while o-toluidine is likely a tobacco-related impurity.^78,79 The measured average quantities for the other six HPHCs are near the method LOQs. For example, acrylamide is quantifiable in eight of the 37 brands examined (22%) with an average quantity of 1.00 ± 0.14 μg/g (LOQ: 0.743 μg/g). The low quantities may imply that they are impurities (e.g., acrylamide,^80,81 arsenic,^82,83 selenium,⁸² catechol⁵⁸) or contaminants (e.g., acrylamide,^80,81 arsenic,⁸⁴ cobalt,²⁶ MEK⁸⁵) related to e-liquid formulation, manufacturing, or storing, as they were reported to be in tobacco leaves, e-liquid solvents, flavorings, cleaning liquid, or local drinking water.

Combining these results with the literature search findings, 36 of the 93 HPHCs were at or above LOQs in ENDS e-liquids (Table 1). The sources of these 36 quantifiable HPHCs fit into one of four categories:

• 1.Added ingredients (nicotine^58,86,87).
• 2.Impurities from e-liquid formulation or plastic device components and parts (anabasine,^86,88 nornicotine,^86,88 tobacco-specific nitrosamines [TSNAs], including 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone [NNK] and n-nitrosonornicotine [NNN],^58,63 benzene,⁸⁹ ethylbenzene,⁸⁹ styrene,⁸⁹ toluene⁸⁹).
• 3.Impurities or other contaminants from e-liquid manufacturing or storage conditions (BAP,^26,29 benz[a]anthracene [BAA],²⁹ benzo[b/k]fluoranthene [BBF/BKF],^26,29 chrysene,^26,29 dibenz[a,h]anthracene [DahA],²⁹ indeno[1,2,3-c,d]pyrene [IcdP],²⁹ naphthalene^26,29).
• 4.A combination of sources, such as being an impurity, a degradant, and an environmental contaminant (all phenols,^29,58 acrylamide,^80,81o-toluidine,^78,79 all carbonyls,^{39,85,90−92} and most metals and metalloids^{26,82−84,93−96}).

These 36 HPHCs may be areas of focus for ENDS manufacturers, academic researchers, and regulators, as they could potentially be present at quantifiable quantities in e-liquids.

This study finds that 54 HPHCs were below LOQ in e-liquids (Table 1). Potential causes for being below LOQ may include the absence of tobacco in e-liquid formulations and the lack of combustion or thermo-degradation in the e-liquid manufacturing and storing process. For example, HPHCs such as nitrosamines other than TSNAs,^97,98 all radioactive metals,⁹⁹⁻¹⁰¹ acetamide,^102,103 aflatoxin B1,^104−106 ammonia,¹⁰⁷ beryllium,¹⁰⁸ caffeic acid,¹⁰⁹ coumarin,^109,110 ethyl carbamate,^111,112 mercury,^113,114 vinyl acetate,¹¹⁵ and chlorinated dioxins and furans¹¹⁶ may be tobacco-related (inherent, agriculture, processing, manufacture, storing) constituents because they are also found quantifiable at low quantities in TCTPs. HPHCs including heterocyclic aromatic amines,^117−120 2,3-benzofuran,¹²¹ furan,¹²² hydrogen cyanide (HCN),^123−125 isoprene,¹²⁶ nitrobenzene,¹²⁷ nitromethane,¹²⁸ and quinoline¹²⁹ are constituents formed from combustion or thermo-degradation. Additionally, HPHCs including many PAHs,^130−133 aromatic amines,⁷⁸ 1,3-butadiene,¹³⁴ 2-nitropropane,¹³⁵ acrylonitrile,¹³⁶ ethylene oxide,¹³⁷ hydrazine,^138−140 propylene oxide,^13,141 and vinyl chloride¹⁴² are from multiple sources (e.g., tobacco agriculture, combustion) that are not expected to affect e-liquid constituents. Therefore, the absence of tobacco or tobacco-related ingredients in e-liquid formulations and the lack of combustion or thermo-degradation in the e-liquid manufacturing or storing process could result in these constituents being below LOQ in e-liquids. All 54 HPHCs that are found below LOQ at this time may warrant further monitoring when new ENDS designs (e.g., formulation, engineering) appear, as more sensitive analytical methods become available, or when new data for products not studied here become available.

The quantifiability of 2,6-dimethylaniline, carbon monoxide (CO), and o-anisidine in e-liquids remains a knowledge gap. Among these three HPHCs, CO is highly volatile and, therefore, unlikely to be present in e-liquids. Without testing the products, the presence of 2,6-dimethylaniline and o-anisidine in ENDS liquids cannot be estimated based on their chemical structures alone, as other aromatic amines were split between the quantifiable and not quantifiable categories for e-liquids. These HPHCs may warrant investigation as accurate, reliable, and fit-for-purpose analytical methods become available.

ENDS Aerosol HPHCs

Among the 67 HPHCs analyzed (64 HPHCs in the study scope +3 additional HPHCs; see Table S1 for inclusion rationale), 14 HPHCs (acetamide, acrylamide, anabasine, caffeic acid, catechol, cobalt, cresols, mercury, n-nitrosodimethylamine [NDMA], nornicotine, o-toluidine, phenol, propylene oxide, styrene) are quantifiable in the aerosols of 3–92% of the 37 brands examined (Table 1, Table S6). These 14 HPHCs comprise 16 measured constituents, as cresols are one HPHC measured by three constituents (i.e., m-cresol, o-cresol, p-cresol). Six of the 16 constituents (anabasine, cobalt, nornicotine, o-cresol, p-cresol, propylene oxide) have averages that are relatively higher than their respective method LOQs. For example, anabasine is quantifiable in the aerosol of 12 of the 37 brands examined (32%) with an average yield of 0.146 ± 0.123 μg/puff (LOQs: 0.00987–0.0444 μg/puff). As discussed for the quantifiable HPHCs in e-liquids, anabasine and nornicotine are known nicotine-related impurities, cresols^29,58,143 are likely impurities from multiple sources, propylene oxide is a PG thermo-degradation-related impurity,¹⁴³ and cobalt may be a contaminant²⁶ from product storage or the sample collection process. The other 10 constituents have measured average quantities near their respective method LOQs. For example, styrene is quantifiable in the aerosol of three of the 37 brands examined (8%) with an average yield of 8.92 ± 1.01 ng/puff (LOQs: 1.11–5.01 ng/puff). The low yields may again imply that they are either tobacco impurities (acetamide,^102,103 acrylamide,¹⁰³ caffeic acid,^109,144 catechol,^143,145,146 cresols,¹⁴³ mercury,^113,114 NDMA,¹⁴⁷o-toluidine,^78,79 phenol,^143,145,146 styrene^89,126,143) or contaminants (acrylamide,⁸¹ catechol,^143,145,146 mercury,^148,149 NDMA,¹⁵⁰o-toluidine^78,79) transferred or heat-generated from formulation, manufacturing, or storing, as they have been reported in tobacco, smoke, e-liquid solvents, or drinking water.

Combining these results with the literature search findings, 34 of the 93 HPHCs were at or above LOQs in ENDS aerosols (Table 1). The sources of these 34 quantifiable HPHCs fit into one of three categories:

• 1.Impurities transferred from e-liquids (alkaloids^151−153).
• 2.Thermo-degradation products from e-liquid ingredients (CO,^7,32,33 propylene oxide^13,141).
• 3.Impurities from multiple possible sources, such as forming from pyrolysis, transferring from e-liquid, leaching from components, or contamination (amides,^{81,102,103,154} aromatic amines,^31,78,79 caffeic acid,^109,155 most carbonyls,^39,91,92 most metals,^95,148 n-nitrosodimethylamine [NDMA],^147,150,156 TSNAs^68,157 BAP,^23,26,29 all phenols,^{29,58,143,145} benzene,^143,158,159 styrene,^89,126,143 toluene^69,89,143).

Most aerosol HPHCs are from a combination of sources, as heating temperature adds potential reaction pathways. These HPHCs may be of strong interest for ENDS manufacturers, academic researchers, and regulators, as they could be present at quantifiable yields in ENDS aerosols that have the potential to reach users. The finding also implies that contamination control may play an important role in reducing the number of HPHCs in ENDS aerosols.

Fifty-five HPHCs were below LOQ in ENDS aerosols (Table 1). Potential causes for being below LOQ may include the absence of tobacco in the product formulations, lack of tobacco combustion, and inability to transfer from liquids to aerosols during product use. For example, nitrosamines (excluding TSNAs and NDMA),^97,98 radioactive metals,⁹⁹⁻¹⁰¹ ammonia,¹⁰⁷ beryllium,¹⁰⁸ ethyl carbamate,^111,112 vinyl acetate,¹¹⁵ and chlorinated dioxins and furans¹¹⁶ may be tobacco-related HPHCs, as they tend to only be quantifiable in TCTPs at low quantities.⁵ HPHCs including 2,6-dimethylaniline,⁷⁸ heterocyclic aromatic amines,^117−120 2,3-benzofuran,¹²¹ furan,¹²² HCN,^123−125 isoprene,¹²⁶ nitrobenzene,¹²⁷ nitromethane,¹²⁸ and quinoline¹²⁹ are constituents generated from tobacco combustions, while many PAHs,^130−133 1,3-butadiene,¹³⁴ 2-nitropropane,¹³⁵ acrylonitrile,¹³⁶ arsenic,⁸²⁻⁸⁴ ethylene oxide,¹³⁷ hydrazine,^138−140 MEK,^5,85,126 selenium,^82,95 and vinyl chloride¹⁴² are from multiple sources that are not expected to affect ENDS constituents (e.g., tobacco agriculture, combustion). Additionally, MEK, arsenic, selenium, BAA, BB/KF, chrysene, DahA, IcdP, naphthalene, and ethylbenzene do not appear to readily transfer from e-liquid to aerosol based on the results of this study. These HPHCs may warrant further monitoring when new ENDS designs appear (especially those using tobacco-derived nicotine, flavors, or extracts), when more sensitive analytical methods become available, or when new data suggest that the puff block approach to measuring an HPHC does not capture the full spectrum of HPHC generation from a product.

Knowledge gaps remain regarding the quantifiability of four HPHCs in aerosols (n-nitrososarcosine [NSAR], aflatoxin B1, coumarin, and o-anisidine). Aflatoxin B1, coumarin, and NSAR are unlikely to be in aerosols, as they were not observed in e-liquids and are not expected to be formed during ENDS heating processes. The quantifiability of o-anisidine is not predictable based on its chemical structure alone, as the literature and this study observe the presence of some, but not all, other aromatic amines in ENDS aerosols. These HPHCs may warrant further investigation when analytical methods that can accurately and reliably determine the HPHCs in ENDS aerosol become available.

Correlations for ENDS HPHCs

Figure 1 is a heatmap that displays the percentage of brands for which HPHCs were quantifiable in this study. For the six HPHCs (anabasine, acrylamide, catechol, cobalt, nornicotine, o-toluidine) quantifiable in both e-liquids and aerosols, the comparison analyses conducted across matrices for 35 of the 37 brands (four products were only measured in one of the two matrices) implies that the HPHCs are either transferrable from e-liquids to aerosols, generated from degradation or heat formation, leached from product components, contaminated from product use or testing, or resulted from a combination of multiple sources. For example, all six HPHCs are quantifiable in both e-liquids and aerosols with overlapping products between the two matrices for anabasine, catechol, nornicotine, and o-toluidine but little to no overlapping products for acrylamide and cobalt. This suggests ready transfer of anabasine, catechol, nornicotine, and o-toluidine from e-liquid to aerosol but potentially low transfer for acrylamide and cobalt. Nornicotine was also quantifiable in the aerosols of 24 products where nornicotine was absent from the e-liquid. As the aerosol nornicotine measurements for all products were performed nine months after the e-liquid measurements, nornicotine may have formed in the e-liquids of these 24 products from nicotine degradation during the nine-month storage at room temperature and then transferred to the corresponding aerosols.^{152,153,160,161} Cobalt was also quantifiable in the aerosols of seven products where cobalt was absent from the e-liquid, implying potential cobalt release from components during product use.⁹⁵ Acrylamide, catechol, and o-toluidine were quantifiable in aerosols of products where they were absent in the corresponding e-liquids, suggesting their potential generation during use from certain product designs (i.e., formulation), in addition to transfer.^{31,58,78−81,103,143,145}

For the other HPHCs in Figure 1, the study results do not show a direct correlation between HPHC quantities in e-liquid and aerosol. For example, arsenic, MEK, and selenium were quantifiable in e-liquids only, suggesting low transfer efficiency for these HPHCs from e-liquids to aerosols. On the other hand, acetamide, caffeic acid, propylene oxide, mercury, and NDMA were quantifiable in aerosols of one or two products only, suggesting that these HPHCs either were generated during product use or were contaminants from product performance or sample analysis. Additionally, this study found that cresols, phenol, and styrene were quantifiable in aerosols from the laboratory testing and quantifiable in e-liquids based on the literature search investigation (this study did not test them in e-liquids). Thus, comparison of the test results and the literature data suggests that they may be transferrable from e-liquid to aerosol.²⁹ For all HPHCs in Figure 1, it may be insightful to further examine the explicit mechanisms of formation, the efficiency of transfer, and the source of contamination for each ENDS matrix. In addition, researchers may further analyze the impact of device type, temperature, flavor, and nicotine type on HPHC formation and transfer.

Comparison to Other Tobacco Products

This study finds that ENDS e-liquids contain a relatively lower number of quantifiable HPHCs (39% with 3% remaining knowledge gaps; Table S2) compared to TCTP fillers⁵ (32–49% with 1–48% remaining knowledge gaps; Table S2). This difference may become larger when the remaining knowledge gaps are addressed. The differences in product designs and product formulations between ENDS and TCTPs may contribute to HPHC quantifiability differences. The HPHCs quantifiable in both TCTPs and e-liquids are nicotine-related HPHCs (alkaloids,^86,88 TSNAs^58,63), common ingredient degradants or impurities (acrylamide,^80,81,162 carbonyls,^{39,91,92,163,164} phenols^{29,58,165−167}), or from a mixture of environmental and product sources (o-toluidine,^78,79 metals and metalloids,^95,96 many PAHs^{26,29,130−133}). The HPHCs that are quantifiable in TCTPs but not in e-liquids are related either to additives in TCTPs (ammonia,¹⁰⁷ coumarin^109,110), to tobacco in TCTPs (acetamide,^102,103 most aromatic amines,⁷⁸ beryllium,¹⁰⁸ caffeic acid,¹⁰⁹ chlorinated dioxins/furans,¹¹⁶ ethyl carbamate,^111,112 mercury,^113,114 all radioactive metals,⁹⁹⁻¹⁰¹ most nitrosamines [excluding TSNAs],^97,98 many PAHs,^130−133 quinoline¹²⁹), or to both tobacco additives and tobacco itself (aflatoxin B1^104−106). Conversely, four HPHCs (benzene, ethylbenzene, styrene, toluene) are quantifiable in e-liquids and absent in TCTP fillers; all four HPHCs are reported to be impurities from common e-liquid ingredients.⁸⁹

The results show that there are fewer HPHCs present at quantifiable yields in ENDS aerosols (37% with 4% remaining knowledge gaps; Table S2) than in TCTP smoke (76–85% with 0–1% remaining knowledge gaps; Table S2).⁵ The HPHCs that overlap between ENDS aerosol and TCTP smoke are nicotine-related (alkaloids, TSNAs),^86,88 thermal degradants (amides,^81,102,154 benzene,^158,159 most carbonyls,^91,164 CO,^7,32,33 NDMA,^147,150,156 phenols,^{145,165−167} propylene oxide,^13,141 toluene^{69,89,126,143}), or from a mixture of environmental and product sources, though these sources often differ between ENDS and TCTPs (most aromatic amines,^31,78,79 B[a]P,^23,130−133 cadmium,^{94,95,156,168} caffeic acid,^109,155 lead,^{95,96,156,168} mercury,^{113,114,148,149} nickel,^95,96,169 styrene^89,126,143). The HPHCs quantifiable in TCTP smoke but not ENDS aerosol can be attributed to two key differences between combustible TCTPs and ENDS: combustion during product use and presence of tobacco. Ammonia,^107,170 2,6-dimethylaniline,⁷⁸ some heterocyclic aromatic amines,^118−120 HCN,^123−125 MEK,¹⁶⁴ most PAHs,^130−133 and most volatile and semivolatiles compounds^{121,122,126,128,129,134,136,137,142,171} are from tobacco combustion. Arsenic,^93,168 chlorinated dioxins/furans,¹¹⁶ ethyl carbamate,^111,112,172 most nitrosamines (excluding TSNAs),⁹⁷ radioactive metals,^99−101,173 selenium,¹⁷⁴ vinyl acetate,¹¹⁵ and vinyl chloride¹⁴² possibly transfer from tobacco. Therefore, these HPHCs are quantifiable in TCTP smoke but not ENDS aerosol. Only two HPHCs (chromium and cobalt) are quantifiable in ENDS aerosol but not TCTP smoke; both HPHCs appear to be linked to ENDS components and parts.⁹⁵

Knowing the HPHC comparability and differences between ENDS (e-liquid, aerosol) and TCTPs (filler, smoke) may inform future risk comparison studies between ENDS and TCTPs and direct the discovery of potentially harmful constituents that are present only in ENDS. The HPHCs that are quantifiable in both ENDS and TCTPs may also be of interest to tobacco product manufacturers, academic researchers, and regulators, who may consider whether these HPHCs may have high potential to be present in new tobacco product designs and matrices.

Our analysis compares tobacco products based solely on whether the HPHC is quantifiable without addressing the amount or yield of each HPHC. These findings do not necessarily mean that ENDS contain fewer harmful constituents than TCTPs. ENDS may contain other harmful constituents that are not captured by the current HPHC list, as observed by the FDA proposed additions to the established 2012 HPHC list in 2019.⁴ New research may further examine the relationship between the quantities of HPHCs and tobacco product types and matrices as well as statistically analyze the data included in Tables S4–S8 for ENDS and reported previously for TCTPs.⁵ Also, our study examined products containing only e-liquids. The HPHCs in and produced by products that contain both e-liquid and tobacco may differ from those included in this study. Further, while the test laboratory was competent and the test methods were fit for purpose, the methods used in this study were not evaluated for whether they were representative of industry or regulatory laboratory test methods. Additionally, this work does not estimate the risk associated with the use of ENDS products as it is outside the scope of this study.

Conclusions

This work addresses knowledge gaps regarding whether an HPHC is quantifiable in ENDS e-liquid and aerosol and how its quantifiability in e-liquid and aerosol compares to each other and to that in TCTP filler and smoke. In conclusion, 36 (39%) and 34 (37%) of the 93 HPHCs on the FDA-established list are quantifiable in ENDS e-liquids and aerosols, respectively, with 25 HPHCs (27%) being quantifiable in both matrices. This work suggests the factors that may influence the HPHC quantifiability in ENDS e-liquids and aerosols are e-liquid formulation, impurities and degradants in the formulations, and contaminants from product manufacturing and storage conditions and environments. Comparisons also suggest that the primary driver of the differences in HPHC quantifiability between TCTPs and ENDS relates to either the presence of tobacco in product formulation or combustion during product use. Product design (product components and parts) also contributes to HPHC differences when HPHCs are present in ENDS and not TCTPs. Future correlation analyses for HPHCs in ENDS versus ENDS device types and other tobacco products may be helpful for understanding comparative product risks and their impact on public health. The results of this study may also be helpful in supporting future hypothesis testing, including mechanistic factors influencing HPHC quantifiability.

Additionally, this study and its findings reflect the diversity of ENDS products on the U.S. market in 2019–2020. The two literature searches conducted for this work indicate that the ENDS product market changes rapidly, with the first search reporting mainly three unique types of ENDS across 15 publications and the second literature search reporting approximately six unique types of ENDS across 41 publications. This study reveals whether the ENDS (e-liquids and aerosols) marketed in 2019–2020 contain quantifiable HPHCs based on a single laboratory testing result across various types of ENDS and does not necessarily predict HPHC quantifiability in future ENDS that have different product designs. As the marketplace changes following evolving ENDS product design, similar studies in the future may be helpful in exploring the quantification of ENDS HPHCs.

The findings of this study can inform the scientific basis for regulators and manufacturers to determine which HPHCs may need to be reported and which specific tobacco product types and matrices may pose increased public health concerns. The study findings also provide information on which HPHCs may be more difficult to report at present. These findings can be helpful for informing public communications regarding the health risks of tobacco products. Lastly, the findings stress the role that contamination control can play in reducing the number of HPHCs in ENDS.

Glossary

Abbreviations

ACM

aerosol collected mass

AαC

2-amino-9H-pyrido[2,3-b]indole

BAA

benz[a]anthracene

BAP

benzo[a]pyrene

BBF

benzo[b]fluoranthene

BCP

benzo[c]phenanthrene

BDL

below the LOD

BJA

benz[j]aceanthrylene

BKF

benzo[k]fluoranthene

BLOQ

below the LOQ

CcdP

cyclopenta[c,d]pyrene

CFP

Cambridge filter pad

CO

carbon monoxide

CT

coil temperature

CV

coefficient of variation

DaeP

dibenzo[a,e]pyrene

DahA

dibenz[a,h]anthracene

DahP

dibenzo[a,h]pyrene

DaiP

dibenzo[a,i]pyrene

DalP

dibenzo[a,l]pyrene

ENDS

electronic nicotine delivery system

EOL

end-of-life

EWL

e-liquid weight loss

FD&C Act

United States Federal Food, Drug, and Cosmetic Act

FDA

United States Food and Drug Administration

Glu-P-1

2-amino-6-methyldipyrido[1,2-a:3′,2′-d]imidazole

Glu-P-2

2-aminodipyrido[1,2-a:3′,2′-d]imidazole

HCN

hydrogen cyanide

HHS

U.S. Department of Health and Human Services

HPHC

harmful and potentially harmful constituents

IcdP

inden[1,2,3-cd]pyrene

IQ

2-amino-3-methylimidazo[4,5-f]quinoline

ISO

International Organization for Standardization

LOD

limit of detection

LOQ

limit of quantification

MeAαC

2-amino-3-methyl-9H-pyrido[2,3-b]indole

MEK

methyl ethyl ketone

NDEA

n-nitrosodiethylamine

NDELA

n-nitrosodiethanolamine

NDMA

n-nitrosodimethylamine

NMEA

n-nitrosomethylethylamine

NMOR

n-nitrosomorpholine

NNK

4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone

NNN

n-nitrosonornicotine

NPIP

n-nitrosopiperidine

NPYR

n-nitrosopyrrolidine

NSAR

n-nitrososarcosine

PAHs

polycyclic aromatic hydrocarbons

PCDDs

polychlorinated dibenzo-p-dioxans

PCDFs

polychlorinated dibenzofurans

PG

propylene glycol

PhIP

2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine

RYOs

roll-your-own tobaccos

SOP

standard operating procedure

STPs

smokeless tobacco products

TCTPs

tobacco-containing tobacco products

Trp-P-1

3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole

Trp-P-2

1-methyl-3-amino-5H-pyrido[4,3-b]indole

TSNAs

tobacco-specific nitrosamines

U.S.

United States

VG

vegetable glycerin

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.chemrestox.4c00093.

• Results of the literature search and HPHC data gap identification in ENDS matrices (Table S1); results of literature search and HPHC data gap identification across tobacco product matrices (Table S2) (PDF)

• Methods (Table S3); product details and end of life (EOL) (Table S4); HPHCs in e-liquids and aerosols (Tables S5 and S6); coil temperature measured by brand (Table S6); and aerosol collected mass (ACM) and e-liquid weight loss (EWL) for HPHCs (Tables S7 and S8) (XLSX)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. CRediT: Samantha M. Reilly formal analysis, investigation, visualization, writing-original draft, writing-review & editing; Tianrong Cheng data curation, formal analysis, investigation, methodology, visualization, writing-original draft, writing-review & editing; Charles Feng conceptualization, methodology, resources, supervision, writing-review & editing; Matthew J. Walters resources, supervision, writing-review & editing.

This work was supported by the Center for Tobacco Products of the U.S. Food and Drug Administration under Contract ID number 75F40119D10003 with Enthalpy Analytical.

The authors declare no competing financial interest.

Notes

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Food and Drug Administration or the U.S. Department of Health and Human Services (HHS). The use of brand names in this publication does not constitute an endorsement by FDA and/or HHS of any product, service, or enterprise.