The Effects of Inhaled Flavors on Intravenous Nicotine

Abstract

Menthol is the only available flavor in combusted tobacco cigarettes; however, e-cigarettes are available in thousands of flavors. Research on flavors and rewarding properties of nicotine is limited. The present study sought to examine the acute rewarding effects of flavors inhaled from an e-cigarette, in combination with intravenous (IV) nicotine among cigarette smokers. In the present study, 24 menthol-preferring young adult (aged 18 to 30) cigarette smokers were tested under 3 different e-cigarette flavor conditions (menthol, green apple, or menthol + green apple) in a within-subject cross-over design. During each test session, each participant received 3 IV infusions (saline, 0.25 mg/70 kg nicotine, 0.5 mg/70 kg nicotine) administered 1 hr apart. The main outcome measures assessed cardiovascular, subjective, and cognitive domains. Compared with green apple or green apple + menthol, menthol produced higher ratings of “cooling” (ps < 0.01). Craving was rated higher following administration of green apple and the combined menthol + apple flavor compared to menthol alone (ps < 0.05). As expected, IV-nicotine dose-dependently increased the ratings of subjective liking/disliking and peak heart rate, improved cognitive performance, and reduced smoking urges (all ps < 0.05). These subjective, cognitive, and physiological effects of nicotine were not affected by any flavor condition. The present findings did not support an interaction between IV-nicotine dose and inhaled flavor for acute effects of nicotine. Green apple flavor, alone or in combination with menthol, could result in higher craving or insufficiently alleviate craving, relative to menthol flavor alone. Additional research is warranted to examine extended exposure to inhaled flavors on the rewarding and addictive effects of nicotine.

With the exception of menthol, all characterizing flavors have been banned from use in combusted tobacco cigarettes (Food & Drug Administration, 2018). However, flavors are not prohibited from use in other tobacco product categories, including e-cigarettes, cigars, and cigarillos, where the majority contain a various flavor additives. Prior studies of tobacco cigarettes have shown that the addition of sweet flavors increased the appeal of these products, especially among youth (Ambrose et al., 2015; King, Tynan, Dube, & Arrazola, 2014). Similar recent data for noncombusted tobacco products has informed the U.S. Food and Drug Administration’s (FDA) prioritization of understanding the impact of flavors on the appeal and use of flavored tobacco products (Wackowski et al., 2018).

Flavors are particularly ubiquitous in e-cigarettes and refill liquids marketed for use in the expanding array of electronic nicotine delivery systems (ENDS), including e-cigarettes. There are thousands of e-cigarette flavor products available on the market, each containing a variety of compounds that are effectively transferred via use of ENDS (Allen et al., 2016; DeVito et al., 2020; Tierney, Karpinski, Brown, Luo, & Pankow, 2016; Zhu et al., 2014). Many observational and epidemiological studies found that e-cigarette users, especially youth, prefer e-liquids with flavors (Pesko, Kenkel, Wang, & Hughes, 2016; Soule, Lopez, Guy, & Cobb, 2016). Fruit and sweet are the most prevalent flavors and, among youth and young adults, flavors as the primary motivation to use e-cigarettes (Harrell et al., 2016; Kong, Morean, Cavallo, Camenga, & Krishnan-Sarin, 2015). Youth and adult ever-users of tobacco and e-cigarettes (aged ≥15) most commonly cited “taste” as the most important reason underlying their choice of their preferred brand of e-cigarettes (Laverty, Vardavas, & Filippidis, 2016). Although the majority of research has focused on the impact of flavors in youth, adults also report a preference for flavors. For example, in a laboratory study wherein adult e-cigarette users self-administered e-cigarettes containing moderate nicotine levels (12 ng/mL) and various flavors, “liking” ratings were positively correlated with “sweet” and “cooling” ratings, but negatively correlated with “bitter” and “harsh” ratings (Kim et al., 2016). These studies suggest that flavors increase the appeal of e-cigarettes and may therefore contribute to initiation and maintenance of e-cigarette use (Ambrose et al., 2015; DeVito & Krishnan-Sarin, 2018).

Menthol is one of the most intensively studied flavors used in tobacco products. It has a well-characterized cooling and soothing action in the airways that may enhance the appeal of menthol cigarettes by reducing the harshness of tobacco smoke (Wise, Breslin, & Dalton, 2012). This action may be particularly significant for youth who are experimenting with combusted tobacco products. For example, data from the National Survey on Drug Use and Health reveal that 44.7% of current smokers between the ages of 12- and 17-years-old smoked menthol cigarettes, compared to 30.1% of adults aged 26 or older (Rock, Davis, Thorne, Asman, & Caraballo, 2010). In addition, numerous cross-sectional studies have shown that menthol cigarette smoking is a risk factor for the development of dependence (Collins & Moolchan, 2006; Hersey et al., 2006; Muscat et al., 2009; Wackowski & Delnevo, 2007). A prospective study of smokers aged 17 or younger demonstrated that smoking initiation with menthol cigarettes was associated with higher risk of progression to established smoking, as well as higher levels of nicotine dependence (Nonnemaker et al., 2012). While these studies indicate that menthol plays a role in both the initiation and maintenance of tobacco product use, the underlying mechanisms by which menthol may facilitate nicotine dependence in humans have not been fully elucidated.

The identification of common mechanisms by which flavors may influence tobacco use and appeal is complicated by the multitude of flavor combinations in tobacco products, particularly in e-liquids. To reduce the complexity of research on flavored tobacco products, there have been attempts to categorize e-liquid flavors. For example, one recent survey suggested that the majority of e-liquid flavors used by consumers can be categorized as either tobacco (23.7%), fruit (20.3%), dessert/sweets (20.7%), or menthol/mint (14.8%) varieties (Yingst, Veldheer, Hammett, Hrabovsky, & Foulds, 2017). It is important to note that most e-liquid flavors were developed as food additives (e.g., watermelon, apple, cherry, cotton candy, or bubble gum) and many are rated by the FDA as generally recognized as safe (GRAS) for oral consumption, but have not been examined for inhalation use (Tierney et al., 2016). For example, it is unknown if menthol is uniquely different than other flavors that have been used in tobacco products or if other flavors have similar effects as menthol. Although many commercially available e-liquid products include a combination of different flavors (e.g., menthol plus fruit or sweet flavors), it is unknown if such flavor combinations may have additive or synergistic rewarding effects.

There has been limited research assessing the impacts of flavors (e.g., delivered via e-cigarettes) on nicotine perception when nicotine is not delivered via a noninhalation route (e.g., intravenously). This is an important avenue of research because the effects of flavors can be separated from other oral and respiratory reinforcers associated with nicotine. For example, in vitro research has shown that menthol can have direct effects on nicotinic receptor functioning (Hans, Wilhelm, & Swandulla, 2012) and preclinical research indicates that that menthol can increase nicotine reinforcement even when not delivered via oral/inhalation route (Biswas et al., 2016). In humans, we have previously used intravenous (IV) nicotine delivery, with or without flavor coadministration, to investigate the impact of menthol delivery via e-cigarette and/or menthol-preference on response to nicotine (DeVito, Valentine, Herman, Jensen, & Sofuoglu, 2016; Valentine, DeVito, Jatlow, Gueorguieva, & Sofuoglu, 2018). Importantly, use of IV nicotine removes any nicotine-associated oral/respiratory tract effects, thus permitting assessment of the reinforcing effects of flavors in the absence of potentially confounding oral/respiratory effects of nicotine.

Current Study

In this study, we compared the effects of flavored e-liquids that contain menthol only, a fruit flavor (green apple) only, and a combination of green apple + menthol, either alone or with IV administered nicotine. Study outcomes included aversive and positive subjective effects, withdrawal severity, urges to smoke, cognitive performance, and cardiovascular measures. We enrolled only menthol-preferring smokers because in our recent study on the acute effects of inhaled menthol on IV nicotine, effects were more prominent in menthol-preferring smokers compared to the non-menthol-preferring sample (Valentine et al., 2018). We chose to use green apple because fruit flavors represent one of the most commonly consumed e-liquid flavors. The use of menthol, green apple, and green apple + menthol flavors enabled the examination of whether these two popular flavors, in combination with nicotine, have synergistic effects using a broad range of outcomes. We hypothesized that the combined menthol and green apple would be more effective than either flavor alone in enhancing the positive subjective effects of nicotine.

Method

Participants

A total of 26 non-treatment-seeking smokers (aged 18 to 30) were recruited from the New Haven, Connecticut area. Potential participants reported daily smoking (i.e., at least one cigarette/day) for the past year, and active smoking status was confirmed by a screening urine cotinine >10 ng/ml (NicAlert). Only menthol cigarette smokers were recruited to minimize the impact of flavor preference on study outcomes and prior e-cigarette experience was not required. Participants were medically healthy and did not meet criteria for current Axis I psychiatric disorders, including alcohol or drug dependence (other than nicotine), as determined by the Structured Clinical Interview for DSM–IV (SCID; First, Spitzer, & Gibbon, 1996) and urine toxicology screening. Participants with a urine cotinine less than 10 ng/ml or positive urine toxicology screen for drugs of abuse (except cannabis) were excluded from participating. Twenty-six participants were eligible for participation with 24 randomized after completion of the adaptation session. Reasons for exclusion after adaptation session included poor venous access (N = 1) and a positive drug screen (N = 1). Nineteen participants completed all three test sessions, one attended two sessions, and four participants attended only the first session. The protocol was approved by the Yale (2000021591) and VA Connecticut Institutional Review Boards (MS054). Written informed consent was provided prior to participation, for which participants were compensated.

Procedure

This outpatient, double-blind, placebo-controlled study consisted of an adaptation session followed by three test sessions. Prior to test sessions, participants were told that they would be inhaling flavors via an e-cigarette and nicotine would be administered via IV immediately thereafter. All participants were randomized to a test session order and received either menthol (2%), green apple (2%), or green apple + menthol (2% menthol + 2% green apple) at each test session, delivered by standardized inhalation from an e-cigarette just prior to each nicotine infusion (a single flavor condition for each test session). Within each test session, all three IV nicotine conditions were tested, 1 hr apart, by delivering saline, lower dose nicotine (0.25 mg nicotine/70 kg), and higher dose nicotine (0.5 mg nicotine/70 kg), in a random order, just after last inhalation. For each participant, the randomized nicotine infusion sequence was fixed across the three test sessions, each performed at least 24 hr apart.

Adaptation session.

To reduce variability in flavor delivery, participants were introduced to the operation of the test e-cigarette that contained a control e-cigarette solution (e-liquid) with tobacco flavor and 0.0% menthol. Participants were coached on inhaling more softly, but for longer (3–4 s) than is typical for combusted cigarettes (Farsalinos, Romagna, Tsiapras, Kyrzopoulos, & Voudris, 2013; Vansickel & Eissenberg, 2013). Participants were instructed to avoid exposure to menthol products (other than their usual brand of cigarettes) for 24 hr prior to each test session, and to abstain from smoking and eating after midnight prior to test sessions. Typical morning caffeine intake was encouraged to avoid withdrawal symptoms that might confound interpretation of study measures.

Test sessions.

Test sessions started between 8 a.m. and 9 a.m. Participants were first evaluated for exclusionary drug use and pregnancy by urine testing, as well as for recent smoking (breath CO <10 ppm; Vitalograph, Inc., Lenexa, KS) and recent alcohol use (breathalyzer, Alco-Sensor IV, Intoximeters, Inc., St. Louis, MO). Participants with a breath CO >10 ppm were rescheduled for another test session. After a light breakfast, an indwelling 20-gauge, flexible catheter with multiple ports was inserted into an antecubital vein of the participants for blood sampling and infusions. Heart rhythm was monitored with a three-lead EKG and blood pressures were acquired using an arm cuff placed opposite to the catheterized arm.

Drugs

Flavor administration.

To ensure appropriate standardization, e-liquids were prepared by Pace Engineering Concepts (Delafield, WI) using commercially available e-liquids that were purchased from the AmericaneLiquidStore™. E-Liquids were nicotine free (0.0% nicotine) in a 1:1 mixture of propylene glycol (PG) and vegetable glycerin (VG). The two menthol-containing e-liquids, menthol and green apple + menthol, had similar menthol concentrations (about 2%) and the green apple flavor had a flavor concentration of approximately 2%. The menthol content and absence of nicotine in the stock e-liquids were verified by the Jatlow Laboratory at Yale School of Medicine. We used Joyetech eGo-C™ e-cigarette configured with a single coil atomizer (2.2 ohm) and a 650-mAH battery operating at 3.7 V (6.2 W). Participants took six inhalations each lasting 3–4 s, one every 15 s, over 90 s just prior to each IV infusion.

Nicotine administration.

Nicotine bitartrate (Interchem Corporation, Paramus, NJ) infusates were prepared the morning of each test session by the West Haven VA research pharmacy and consisted of either physiological saline, or nicotine dissolved in saline (at 0.25 and 0.5 mg nicotine/70 kg bodyweight) in a volume of 5 ml. Just after the last of six inhalations, a 30-s infusion was administered by a study physician, with each infusion given 1 hr apart. In our prior research, these nicotine doses have been shown to be well tolerated, produce expected physiological responses, and both positive and negative subjective effects (Valentine et al., 2018). In order to allow subjective responses to return to baseline, the infusions were given 1 hr apart (Sofuoglu, Herman, Nadim, & Jatlow, 2012; Sofuoglu, Mouratidis, Yoo, Culligan, & Kosten, 2005; Sofuoglu, Yoo, Hill, & Mooney, 2008; Valentine et al., 2018).

Outcome Measures

Outcome measures were administered at multiple time points within each session. Baseline measures were administered 5 min prior to the first infusion of the session. Other measurements occurred at time intervals following each infusion (i.e., saline, 0.25 mg nicotine/70 kg, and 0.5 mg nicotine/70 kg). Finally, postsession measures were administered at the end of each session (i.e., 180 min after initial infusion).

Cardiovascular.

Cardiovascular measurements (heart rate, systolic and diastolic blood pressure) were collected at baseline and at 1, 2, 3, 5, 8, 10, and 15 min following each IV infusion. Peak changes in systolic blood pressure (SBP), diastolic blood pressure (DBP) and heart rate (HR) were calculated as the maximum achieved value during each infusion period minus the baseline value that was obtained just prior to each infusion.

Subjective.

Measures included the Drug Effects Questionnaire (DEQ), the Minnesota Nicotine Withdrawal Scale (MNWS), and the Brief Questionnaire of Smoking Urges (BQSU). The DEQ captured acute drug effects by recording participants’ response intensity on a 100 mm visual analogue scale ranging from not at all to extremely. The DEQ consisted of 11 items: cooling effect, dislike the sensation, any sensations (in mouth, throat or chest), feel a drug effect, high, feel stimulated, feel a head rush, like drug effect, dislike any effects, craving a cigarette, and would like more of the drug. The DEQ was administered at baseline and then at 1, 3, 5, 8, 10, 15, 30, 45, and 55 min after each infusion. The MNWS is a widely used eight-item scale measuring symptoms of tobacco withdrawal (Hughes & Hatsukami, 1986). A modified MNWS total score was used by summing Items 1–7 and removing Item 8 (“I have difficulty sleeping”), because Item 8 would not be expected to be sensitive to changes within the test session. The MNWS was administered at baseline and postsession. The BQSU is a 10-item scale with two factors rated on a 7-point Likert scale: Factor 1 reflects urges to smoke for stimulation, and Factor 2 reflects urges to smoke to relieve negative mood and withdrawal (Cox, Tiffany, & Christen, 2001). The BQSU was administered at baseline and 55 min after each IV infusion.

Cognitive.

The Automated Neuropsychological Assessment Metrics (ANAM) Version 4 was used to administer three cognitive tasks: continuous performance task (CPT), mathematical processing task (MPT), and Stroop. Similar cognitive tasks have exhibited sensitivity to nicotine withdrawal and administration (Myers, Taylor, Moolchan, & Heishman, 2008). All cognitive tasks were administered at baseline and 15 min after each infusion. The ANAM “throughput” score (a measure of both accuracy and speed; higher scores reflect better task performance) was used as the outcome measures for each cognitive task (CPT, MPT, Stroop; Thorne, 2006). The CPT assesses sustained attention and working memory. Participants were instructed to press one of two buttons to indicate whether a letter was the same as the previous presented letter. The MPT assesses basic computational skills, attention, and working memory. Participants were instructed to press one of two buttons to indicate whether a three-integer equation (e.g., 5 + 3–1) was greater or less than 5. The Stroop assesses cognitive control and consisted of three stimulus levels. In the first level (i.e., word condition), participants were instructed to press one of three colored buttons in response to a color word (e.g., “blue”) presented in white ink. The second level (i.e., color condition) instructed participants to press the corresponding colored button in response to “XXX” presented in one of three ink colors. Finally, the third level (i.e., interference condition) contains words displayed in incongruent ink colors (e.g., “blue” in yellow ink). Participants were instructed to press the button corresponding to the ink color (which requires overriding the automatic response to read the word). Only responses from the third level were used to calculate a Stroop “throughput” score.

Data Analyses

For each participant, peak values up to 60 min after each infusion in Sessions 1, 2, and 3 were extracted for the following DEQ items: “feel a drug effect,” “cooling effect,” “like drug effect,” “dislike any effects,” and “craving a cigarette” items were analyzed separately. These items were selected a priori based on the expected effects of flavors (i.e., inducing cooling, reducing aversive effects of nicotine, increasing drug liking or craving). Cardiovascular and subjective measures were assessed for normality. For DEQ measures that were skewed, square root transformations were used to bring the variables more in line with the normal distribution.

A series of mixed effect models were used that are designed to handle data nested within participants and allow for different number of observations per participant (e.g., due to attrition). The model for DEQ items included the within-subject effects of flavor (menthol, green apple, combination [menthol + green apple]) and nicotine dose (saline, lower, higher), and all possible interactions. Session (test Session 1, 2, or 3; flavor condition randomized across sessions) and period (order of nicotine or saline conditions within sessions) were also added to the models to account for period effects in menthol and nicotine administrations, respectively. Random effects for subject, nicotine dose, and flavor within subject were used to model the correlations among repeated measures on the same individual. This was the best-fitting structure identified based on Akaike’s and Schwartz-Bayesian information criteria (AIC and BIC, respectively) for all outcomes. Least square means and standard errors and post hoc comparisons of least square means were used to describe the significant effects for each outcome.

A similar mixed effects model evaluating MNWS and BQSU with flavor and timepoint as within-subject factors was fit. Session was also included in the model. A combination of a random subject effect and compound symmetry structure within session provided the best fit to the data according to BIC. For the BQSU, subscales (Factor 1, Factor 2) were analyzed using mixed effects models with flavor and nicotine infusion (baseline, 55 min after each infusion [three in total per session], and postinfusion) as within-subject factors. A combination of a random subject effect and unstructured variance-covariance of the errors within session provided best fit to the data according to BIC.

For cognitive measures, throughput scores for all three cognitive measures (CPT, MPT, Stroop) were analyzed using separate mixed effects models with flavor and nicotine infusion (baseline and 15 min after each infusion [three in total per session]) as within-subject factors. Interaction terms for nicotine and flavor and session were also included in the model. A combination of a random subject effect and unstructured variance-covariance of the model errors within session provided best fit to the CPT and MPT data, whereas a combination of a random subject effects and compound symmetry of the model errors within session provided best fit to the Stroop data according to the BIC.

Results

Descriptive Statistics and Power

Baseline demographics and tobacco use for study participants (N = 24) measures are presented in Table 1. Means and standard deviations of all study outcome variables are included in online supplementary materials. The study was designed to have 80% power to detect large effect sizes (d′= 0.8) for interactions in repeated measures analysis for 30 participants at 0.05 significance level and correcting for multiple comparisons. Although our sample was below the target N, we were still powered to detect large effect sizes for main effects and interactions and their corresponding least square mean comparisons.

Table 1.

Descriptive Statistics

Variable	M (SD)	n (%)
Demographics
Age (years)	26.2 (2.4)
Sex
Male		15 (62.5)
Female		9 (37.5)
Race/ethnicity
African American		15 (62.5)
Caucasian		7 (29.2)
Other		4 (16.6)
Hispanic ethnicity		6 (25.0)
Smoking severity and tobacco product use
FTND	3.2 (2.1)
Average cigarette consumption per day	8.3 (4.5)
Years of smoking	13.2 (2.4)
Serum cotinine levela	194.4 (123.8)
Positive THC at screen (%)		15 (62.5)

Note. FTND = Fagerstrom Test of Nicotine Dependence; CO = carbon monoxide; THC = tetrahydrocannabinol.

^aCotinine extracted from blood serum samples taken immediately before each test session.

Cardiovascular Effects

Changes in HR were dose-dependent on nicotine, F(2, 132) =24.47, p < .0001; higher dose > lower dose > saline (ps > .05; see Figure 1). There was no significant main effect of flavor or nicotine by flavor interaction term (ps > .18). For DBP and SBP, there were no significant main effects of nicotine, flavor, or interaction terms (ps > .08).

Figure 1.

Main effects of nicotine on cardiovascular and subjective measures. Peak change scores for heart rate (HR) and peak response on Drug Effects Questionnaire (DEQ) ratings of cool, dislike drug, and like drug across three flavor conditions (menthol, green apple, and green apple + menthol). For HR, the columns represent the peak postdose minus predose values with SEM error bars. Values for dislike drug were log transformed prior to analysis to address normality. For DEQ questions, bars represent the average highest reported value up to 60 min after each infusion. * p < .05. ** p < .01. *** p < .001.

Subjective Effects

Ratings for craving for cigarettes were dependent on flavor, F(2, 147) = 4.39, p = .014 (see Figure 2). Pairwise comparisons revealed that the green apple (p = .009) and combination (p = .016) flavors were associated with higher craving than the menthol flavor. There was no difference in rating of craving between the green apple and combination flavors (p = .80). There was no significant main effect of nicotine or interaction term (ps > .17). Ratings for like the drug effects were nicotine dose-dependent, F(2, 147) = 15.08, p < .0001; higher dose > lower dose > saline (ps > .05; see Figure 1). There was no significant main effect of flavor or interaction term (ps > .38). Ratings for dislike the drug effects showed a main effect of nicotine dose, F(2, 147) = 7.81, p = .0001 (see Figure 1). Pairwise comparisons revealed that high nicotine dose was associated with lower rating of dislike drug than low dose (p = .007) and saline (p = .0002). There was no difference in rating of dislike drug between the lower dose and saline (p = .25). There was no significant main effect of flavor or interaction term (ps > .49). Rating of the cooling effects showed a main effect of flavor, F(2, 147) = 9.74, p < .0001 (see Figure 2) and nicotine, F(2, 147) = 5.27, p = .006 (see Figure 1). Pairwise comparisons revealed that menthol flavor was associated with higher rating of cooling than the green apple (p < .0001) and combination (p = .002) flavors. There was no difference in rating of cooling between the green apple and combination flavor (p = .31). Pairwise comparisons revealed that higher nicotine dose was associated with higher rating of cooling than lower dose (p = .03) and saline (p = .002). There was no difference in ratings of cool between the lower nicotine dose and saline (p = .33). The interaction term was not significant (p = .32).

Figure 2.

Main effects of flavor on subjective measures. Peak response on Drug Effects Questionnaire (DEQ) ratings of cool and craving across three nicotine conditions (saline, lower and higher dose). Values for craving were log transformed prior to analysis to address normality. For DEQ questions, bars represent the average highest reported value up to 60 min after each infusion. * p < .05. ** p < .01. *** p < .001.

For the BQSU Factor 1 subscale, there was only a significant main effect of nicotine, F(4, 264) = 4.34, p = .002. Pairwise comparisons revealed that the higher nicotine condition was associated with significantly lower scores than placebo (p = .02), baseline (p = .0001), and postsession (p = .02) measures and low nicotine condition was associated with significantly lower scores than baseline (p = .02). There were no significant effects for the BQSU Factor 2 subscale (ps > .34).

For the MNWS, there was a significant main effect of timepoint, F(1, 91) = 8.16, p = .005; specifically, scores were higher at baseline than at postsession (p = .005). There was no main effect of flavor (p = .45), session (p = .11), or interaction between flavor and timepoint (p = .39).

Cognitive Effects

On the CPT, there was a main effect of nicotine, F(3, 211) =6.46, p = .0003. Pairwise comparisons revealed that CPT Throughput scores were significantly higher (better task performance) on higher and lower nicotine compared to baseline and to placebo. On the MPT, there was a main effect of session, F(3, 198) = 5.70, p = .004. Pairwise comparisons revealed that MPT throughput scores increased substantially from the first to the second and third sessions. On the Stroop, there were significant main effects of nicotine, F(3, 204) = 4.54, p = .004, and of session, F(2, 204) = 3.74, p = .03. Pairwise comparisons revealed that Stroop throughput scores were significantly higher (better performance) on higher nicotine compared with baseline, and on lower nicotine compared to baseline and placebo. Stroop throughput scores also increased substantially from the first to the third sessions. For all cognitive tests, there was no main effect of flavor (ps > .20) or interaction between flavor and nicotine condition (ps > .24).

Discussion

The primary results from this study demonstrate that the green apple + menthol, compared with green apple or menthol flavor, did not enhance or attenuate the liking or disliking of IV-infused nicotine. Second, acute craving was higher in sessions where participants inhaled green apple + menthol or green apple flavor, compared with menthol only flavor. However, flavor had no pervasive impact on smoking urges or withdrawal intensity measured after each infusion and postsession using the BQSU and MNWS, respectively. Third, we found expected effects of nicotine including improved cognitive performance and increased heart rate, but flavor had no effect on these outcomes. These findings do not support our hypotheses that the combined menthol and green apple flavor will be more effective than either flavor alone in enhancing the positive effects of nicotine.

Clinical studies evaluating the impact of flavors on the rewarding effects of nicotine have reported conflicting results. For example, Audrain-McGovern, Strasser, and Wileyto (2016) reported that coadministering nicotine and green apple or chocolate flavoring, compared with unflavored, enhances subjective reward and satisfaction from e-cigarette use (Audrain-McGovern et al., 2016). Another study asked participants to rate preference for five e-liquid flavors coadministered with nicotine (18 mg/ml) and then randomly assigned participants to take home an e-cigarette that varied by nicotine (0, 18 mg/ml) and flavor (preferred flavor or tobacco flavor; Litt, Duffy, & Oncken, 2016). After 6 weeks of use, individuals randomized to cherry or tobacco flavor had the highest frequency of e-cigarette use, and those who received preferred menthol flavor showed the greatest reduction in combusted cigarette use (Litt et al., 2016) suggesting differential effects of flavors on smoking behavior. Conversely, a recent study found negligible effects of three flavors (i.e., cream, tropical fruit, tobacco/menthol) on subjective effects of nicotine (36 mg/ml) when codelivered via e-cigarette (Cobb et al., 2019). In a previous study using a similar methodology we found that inhaled menthol, compared with tobacco flavor, did not change the positive subjective effects from IV nicotine (Valentine et al., 2018). Taken together, flavors in e-cigarettes may not solely motivate e-cigarette use or add to the primary rewarding effects of nicotine, but rather the interaction between flavors and the rewarding effects of nicotine are more nuanced and may be dependent on flavor type and flavor preference of the individual.

One possible mechanism by which flavors could increase the appeal of nicotine-containing tobacco products is by masking or counteracting the aversive effects associated with higher doses of nicotine that occur in the mouth and respiratory tract during inhaled delivery. For example, menthol and nicotine were each rated as causing irritation/harshness at higher doses when administered separately via e-cigarette, but when higher dose menthol and nicotine were coadministered via e-cigarette, irritation/harshness ratings were reduced (Rosbrook & Green, 2016). In addition, in adults, fruit (green apple) and menthol flavors had dissociable effects on e-cigarette appeal in the presence and absence of high nicotine (DeVito et al., 2020). Namely, while both fruit and menthol flavors were rated higher than unflavored on some rewarding properties, particularly in the absence of nicotine (with fruit preferred over menthol in the absence of nicotine), only menthol (not fruit) diminished the aversiveness of high dose nicotine, when coadministered via e-cigarette (DeVito et al., 2020). In the current study, flavor appeared to only have an effect on peak reported craving. Given all participants were menthol-preferring, the lower peak craving reported in the menthol, relative to the green apple or combined conditions, may reflect a familiarity with menthol flavor or abstaining from smoking usual menthol cigarettes and subsequent decrease in craving. Additionally, IV nicotine may have different reward properties compared to coadministration of flavor and nicotine via oral inhalation.

Delivery of nicotine IV is a well-validated approach that has several distinct strengths, relative to inhaled route. IV nicotine eliminates oral and respiratory tract effects (e.g., harshness or bitterness) of nicotine and can contribute to greater blinding in placebo-controlled designs. In animal and human research studies of drug reinforcement, IV self-administration is the gold standard to evaluate dose-dependent effects (Goodwin, Hiranita, & Paule, 2015). IV nicotine also allows for precise dosing that produces similar arterial and venous concentrations that occur from smoking (Rose, Behm, Westman, & Coleman, 1999). Additionally, IV nicotine produces rapid subjective effects, including like the drug and dislike the drug effects, that are similar to inhaled nicotine (Harvey et al., 2004; Jensen, DeVito, & Sofuoglu, 2016; Mello, Peltier, & Duncanson, 2013; Sofuoglu et al., 2008). In the present study, IV nicotine produced expected dose-dependent increases in heart rate, improved cognitive performance, and enhanced ratings of like the drug effects and dislike drug effects. Importantly, the use of IV nicotine and flavor only e-cigarettes enables the investigation of systemic or central effects of nicotine and nicotine-flavor interactions without confounding from the potent oral/respiratory tract effects of nicotine. Other smoking-related cues, such as oral sensory experience of inhaling e-liquid, visual cue of vapor (mimicking cigarette smoke), and behavior of lifting e-cigarette to mouth to puff, were still present; however, these cues are present in all flavor and nicotine conditions so any generalized effects of these cues should not be driving any specific flavor or nicotine level findings. Although the interactions between rewarding effects of nicotine and flavor were largely absent, our results demonstrated greater reported cooling in response to menthol flavor (vs. combined and apple) and high nicotine (vs. saline and low nicotine).

By using an e-cigarette to deliver flavors, local sensory cues of the flavors were maintained but dissociated from the chemosensory cues provided by nicotine. The current model was optimized to examine the effects of different flavors and nicotine, alone or in combination, on multiple outcome measures, while removing the potent effects of nicotine on oral and respiratory tract regions which can be aversive (e.g., harshness, bitterness) but also serve as conditioned cues (e.g., “throat hit”). However, in menthol preferring smokers, the oral sensory cues associated with menthol may serve as conditioned cues paired with nicotine delivery. At characterizing levels, menthol has a “throat hit” of its own which may partially mask nicotine specific “throat hit.” This complex relationship may account for the observation that while nicotine and menthol each have their own harshness, these effects are not additive, but rather menthol can reduce the perceived harshness of nicotine (e.g., DeVito et al., 2020). Prior research has shown that menthol-preferring smokers, compared with non-menthol-preferring smokers, show less alleviation of smoking urges following delivery of IV nicotine (DeVito et al., 2016). In that study, no menthol was delivered and there was no e-cigarette component (i.e., no flavor or other smoking-related cues). One interpretation of those findings was that the menthol may be such a potent smoking-related cue that nicotine in the absence of menthol is less able to satisfy smoking urges in these menthol-preferring smokers (DeVito et al., 2016). That explanation would be consistent with the current study finding that craving was rated as lower following menthol (relative to other flavor conditions) in this sample which was restricted to menthol-preferring smokers.

There are several limitations of the current results. First, recruitment was limited to young adults aged 18 to 30, which may limit the generalizability to other age groups like adolescent and middle age smokers. Second, our sample is relatively small and, although the within-subject study design was intended to maximize power, it is possible that we were unable to detect significant interactions between outcome variables due to low power. Our sample also did not have a sufficient number of female participants to evaluate sex differences in study outcomes. Given the extensive literature on sex differences in nicotine reward and reinforcement (Perkins, Donny, & Caggiula, 1999; Perkins et al., 2006), including in response to IV nicotine (DeVito, Herman, Waters, Valentine, & Sofuoglu, 2014; Jensen, DeVito, Valentine, Gueorguieva, & Sofuoglu, 2016), additional research is needed to evaluate the role of sex and interaction between e-cigarette flavor and nicotine. All participants were menthol-preferring cigarette smokers; considering previously demonstrated group differences in response to IV nicotine (DeVito et al., 2016) and the interactions between IV nicotine and menthol delivered via e-cigarettes (Valentine et al., 2018), the findings may not apply equally to nonmenthol preferring smokers. We did not include a tobacco or unflavored control condition that could elucidate the effect of any flavor (menthol, green apple, or green apple + menthol) relative to nonflavored (or nontobacco flavored) control. Finally, experience with e-cigarette use was not a requirement of the study and regular users of e-cigarettes may respond differently to flavors and IV nicotine. In particular, individuals likely select e-cigarette flavors based on personal preference and appeal. A flavor that is personally appealing may increase nicotine-related reward. Although all participants in the current were menthol preferring, future research would benefit from examining the impact of the effects of other preferred flavors (e.g., other fruit or dessert flavors) on the rewarding effects of nicotine.

There is increased public health interest in the sale and use of e-cigarettes and flavors potentially enhancing the rewarding effects of nicotine. The regulatory landscape for e-cigarette flavors is rapidly evolving and implementation differs on the federal, state, and city levels. Menthol is the only flavor still allowed in combusted cigarettes at characterizing levels and many proposed e-cigarette flavor regulations also propose exempting menthol from regulation in e-cigarettes. The most likely category of flavors to be regulated in e-cigarettes are the nontobacco and nonmenthol flavors (e.g., fruit, etc.) which are already banned at characterizing levels in combusted cigarettes. Indeed, as of February 2020 a federal ban on the sale of nontobacco and nonmenthol flavors in closed cartridge and/or pod systems (e.g., JUUL) went into effect, but these flavors remain available for other forms of e-cigarette devices (Food & Drug Administration, 2020). Given this shifting regulatory landscape, it is important to identify shared and distinct effects of these flavor categories on nicotine appeal. The flavors examined in the current study do not enhance the subjective effects of IV-delivered nicotine in young adults. However, flavors are clearly a motivator in the initial and potential continued use of e-cigarettes, when flavors and nicotine are coadministered via e-cigarettes. The diversity of available flavors leaves open the possibility that other flavor combinations may increase the rewarding effects of IV-delivered nicotine. Future studies can utilize the current study design to evaluate the effect of other flavors and flavor preference on study outcomes.

Public Health Significance.

This study examined whether flavors inhaled through an e-cigarette would enhance the acute rewarding effects of nicotine administered intravenously. Across multiple domains, there was an expected dose dependent response to nicotine; however, flavor had minimal to no effect on the acute effects of nicotine. Any potential enhancement of nicotine via e-cigarette flavors is likely complex and may depend on a variety of characteristics including flavor preference and e-cigarette experience.