The data included in this analysis were from two clinical studies (Study A and Study B) with similar design of randomized, controlled and forced switching. Males and females, 21 years or older, in generally good health as self-reported, smoking 10 or more cigarettes per day were recruited. Interested adult smokers gave written informed consent before enrolling in the studies and were paid for their participation. Participants were free to stop smoking and continue or discontinue the studies at any time for any reason. During the entire study period, participants smoked in controlled clinical settings. In both studies at baseline, participants smoked their regular combustible cigarettes. During the post-baseline phases, the participants were switched to different groups that either vaped the JUUL ENDS, continued smoking their own cigarettes or stopped smoking. Additional details about the two studies have been published previously ( [21] [22] ).

The four JUUL ENDS evaluated in Study A were closed systems consisting of the rechargeable device and disposable pods pre-filled with 0.7 mL of 5% nicotine-salt solution by weight (0.77 g of e-liquid per pod, 59 mg/mL of nicotine, i.e., 40 mg nicotine per pod). The products comprised four commercially available flavors included Virginia Tobacco, Mint, Mango, and Creme. In Study B, there were eight JUUL ENDS studied: Virginia Tobacco, Menthol, Mint, Mango at 3% and 5% nicotine concentrations, respectively. There were three groups evaluated: one Dual Use group that used JUUL ENDS concurrently with up to half of their usual brand (UB) cigarettes smoked per day at baseline; one UB cigarette group continued exclusively smoking their UB cigarettes, and one group abstained from all tobacco and nicotine product use (Abstinence group).

In both Study A and Study B, the biomarkers of exposure included 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL), 3-hydroxypropyl mercapturic acid (3-HPMA), monohydroxy butenyl mercapturic acid (MHBMA), 2-cyanoethylmercapturic acid (CEMA), 1-hydroxypyrene (1-OHP), S-phenylmercapturic acid (S-PMA), and N-nitrosonornicotine (NNN) in urine samples and carboxyhemoglobin (COHb) in blood samples. Nicotine equivalents (nicotine, cotinine, trans-3’hydroxycotinine, and glucuronides) were also collected in urine. In Study A, the urinary biomarkers were measured at baseline and on Day 5 and the blood biomarker was measured daily. In Study B, the urinary and blood biomarkers were measured at baseline and on Day 6. Ortho-toluidine (O-Tol), 2-aminonaphthalene (2-NA), and 4-ABP (4-aminobiphenyl) were measured in Study B only.

Twenty-four-hour urine samples were collected for the determination of urinary biomarker levels. Blood samples were collected for determination of COHb. These biomarkers were measured using validated methods as reported previously ( [8] [10] [11] [14] [15] [23] [24] [25] ). The analysis of the samples for the biomarkers was conducted by the same analytical methods and the same laboratories. The validation criteria used for bioanalytical methods ensured that there were no systematic deviations in the analytical measurement. Nicotine equivalents (NE), a particulate phase (PP) biomarker for nicotine exposure, was calculated as the molar sum of nicotine and its 5 major metabolites in urine, nicotine-N-glucuronide, cotinine, cotinine-N-glucuronide, trans-3-hydroxycotinine, and trans-3-hydroxycotinine-O-glucuronide ( [8] [23] [26] [27] ). Total NNAL, a PP biomarker frequently used to estimate exposure to NNK, a tobacco-specific nitrosamine [5] , was calculated as the molar sum of free NNAL and its conjugated metabolites in urine.

1-OHP, a PP biomarker, often used as a surrogate for polyaromatic hydrocarbon (PAH) exposure in urine ( [5] [28] [29] ), was obtained by measuring free and conjugated metabolites of pyrene. COHb, a GVP biomarker ( [10] [26] [30] ), was used to estimate carbon monoxide exposure. S-PMA, a GVP biomarker, a metabolite of benzene widely used in environmental, occupational and smoking-related exposure studies ( [5] [29] ), estimated benzene exposure; 3-HPMA, a GVP biomarker for smoking-related acrolein exposure ( [10] [31] ) and MHBMA, a GVP biomarker of exposure to 1,3-butadiene [15] , CEMA, a biomarker of acrylonitrile [32] , 2NA, a biomarker for polycyclic aromatic hydrocarbons (PAHs) [33] , and O-Tol, a biomarker of ortho-toluidine [34] , were all analyzed with validated methods.

Descriptive statistics were calculated for the demographic data by study and actual product use. For both combustible cigarettes and JUUL ENDS, the coefficient of variation (CV) for each biomarker was calculated as the standard deviation divided by the mean and multiplied by 100%. Pearson correlation analysis (SAS® PROC CORR) was used to examine the correlation between the biomarkers. SAS® version 9.4 (SAS Institute, Inc., Cary, North Carolina, USA) was used to perform the statistical analysis.

Mean age, BMI, smoking duration in years, and number of cigarettes smoked per day were comparable (P > 0.05) between the two studies. There was a slightly higher proportion of male participants (61.7% vs 51.8%) and a lower proportion of African American JUUL ENDS users (15.0% vs 43.2%) in Study A than in Study B (Table 1).

In the JUUL ENDS, the levels of most biomarkers decreased considerably (90% or greater) from baseline when the participants smoked cigarettes to the last study day when the participants used the JUUL ENDS ( [21] [22] ). In both studies, the coefficient of variation increased from cigarette smoking to JUUL ENDS use in total NNAL, nicotine equivalents, 1-OHP, and CEMA, but decreased in 3-HPMA, MHBMA, and S-PMA. In study A, the average coefficient of variation was 44.76% in cigarette smoking and was 40.71% in JUUL ENDS use. In Study B, the average coefficient of variation was 56.12% in cigarette smoking and was 67.28% in JUUL ENDS use (Table 2(a) and Table 2(b)). NNAL, MHBMA and S-PMA were generally more variable than other biomarkers.

CC: Combustible Cigarette smokers.

The average correlation coefficient between any two biomarkers in Study A was 0.59 at baseline when the participants smoked cigarettes and was 0.11 after the participants switched to JUUL ENDS. This was a considerable decrease (Table 3(a) and Table 3(b)). A similar trend was observed in Study B in which the average correlation coefficient decreased from 0.57 for cigarette smoking to 0.14

for JUUL ENDS use. In both studies, there was a moderate (r = 0.45 or higher) to a high correlation (0.7 or higher) between nicotine equivalents with other biomarkers when the participants were smoking cigarettes (baseline). After the participants switched to JUUL ENDS, the average correlation between nicotine with other biomarkers (not including NNAL) decreased from 0.58 to 0.10 in Study A and from 0.65 to 0.13 in Study B. A similar trend was observed in the correlation between NNAL and other biomarkers. The correlation between nicotine and NNAL did not change much from cigarette smoking to JUUL ENDS use and a linear trend is observed (Figure 1 and Figure 2).

In Study B, the average level of nicotine equivalents was around 40% higher in the 5% nicotine JUUL ENDS than in the 3% nicotine JUUL ENDS, but the levels of other biomarkers were comparable between the two JUUL ENDS groups (Table 4). Levels of NNAL, 3-HPMA, COHb, and HMPMA were less than 5% higher in the 5% nicotine JUUL ENDS and levels of MHBMA, S-PMA, and CEMA were lower. Except for 1-OHP, an increase of nicotine in the JUUL ENDS products did not result in a substantial increase in the levels of the biomarkers in the users. Because the average coefficient of variation and the average correlation coefficient were similar (Table 5) between the 3% JUUL ENDS and the 5% JUUL ENDS, an increase in the nicotine level in the JUUL ENDS also did not change the variabilities of the biomarkers and the correlation between the biomarkers.