To validate and evaluate our PBPK model’s capability for predicting nicotine distribution in the blood, we simulated five different exposure scenarios and compared the outcomes of modeling simulations with published results from clinical studies. As a simple example of the PBPK model application, we considered nicotine PK profile from intravenous administration of nicotine [22]. This example does not need the use of dosimetry and permeation models described earlier, as nicotine is directly injected into the arterial blood. The model prediction of the time-course arterial and venous plasma concentration for nicotine and cotinine during and after 30 min intravenous infusion of nicotine at a rate of 2 µg/kg/min [22] are shown in Fig. 11a–b. Overall, the model provides excellent agreement with both the venous and arterial nicotine concentrations. Given individual physiological and metabolic variability, the agreement for cotinine is good, with model predictions being within ± 20% of the reported data. The model also captured the change in heart rate during and after exposure to nicotine (Fig. 11c), which was also in good agreement with measured outcomes. The overall excellent agreement between experimental and simulated results, speak to the validity of our predictions from the PBPK model for a well-defined exposure scenario. Next, we applied the model for four categories of products as described below.
Conventional cigarettes
To further demonstrate validity of our PBPK model to accurately predict the time course of nicotine PK profile following smoking of conventional cigarettes, we simulated and compared model predictions against measured nicotine PK profiles from the four published studies described below. The studies had varying use conditions (e.g., single vs. repeated use) and products with different nicotine deliveries.
Picavet et al. [23] reported the time-course venous concentration of nicotine in 28 subjects after a single use of a conventional cigarette. Venous plasma was collected at 2, 4, 6, 8, 10, 15, 30, 45, and 60 min, and at 3, 4, 6, 9, 12 and 24 h. In order to simulate the Picavet data, it was necessary to approximate the mass of nicotine inhaled from the conventional cigarette as this was not explicitly reported. We assumed 1.1 mg of nicotine was inhaled from each cigarette in 10 puffs over 9 min.
In another study by Benowitz and coworkers [24], 10 healthy men, 24 to 61 years of age, who were habitual cigarette smokers, were asked to smoke a single cigarette under controlled conditions. Subjects were directed to smoke 1 puff every 45 s, for a total of 12 puffs over 9 min. It was noted that the subjects’ own brand of cigarettes yielded an average of 1.1 mg of nicotine/stick with participants smoking approximately one and one third cigarette over the 9 min.
Two additional studies reporting venous plasma concentrations during and after repeated use of cigarettes were published by deBethizy and coworkers [25] and Russell and coworkers [26]. deBethizy reported the venous plasma concentration for each of the 10 subjects, at various time points during and after smoking 7 cigarettes. The subjects were instructed to smoke at a rate of 1 cigarette every 30 min. The reported nicotine yield was 0.81 ± 0.2 mg per stick. Venous plasma concentrations were reported at 5.5, 7.5, 15, and 30 min after lighting cigarettes 1–6. Venous plasma concentrations were measured and reported at 5.5, 7.5, 10, 15, 20, 25, 30, 45, 60, 120, 180, and 240 min, after lighting of the seventh cigarette. The Russell study [27] reported venous plasma nicotine concentrations in a single subject during and after cigarette use, at a rate of 1 cigarette/h over 7 h. Cigarettes were smoked over a 5 min period at 0, 7, 15, and 30 min after lighting each cigarette. In a second study design, a subject was tasked with smoking 3 cigarettes/h over 7 h. Samples were collected every 20 min starting with the lighting of the first cigarette.
For dosimetry estimates, for all the cigarette smoking scenarios modeled, it was assumed that approximately 95% of the inhaled nicotine mass reaches the alveolar region, where absorption would be instantaneous. The remaining inhaled mass was simulated as being deposited equally in the BC and the URT, with negligible transfer to the GI region. Figure 7c shows instantaneous transfer rate in the LRT. We assumed all the inhaled nicotine from cigarettes is absorbed with no exhaled amount.
The simulation of nicotine venous plasma concentration during and after single and repeated cigarette uses are shown in Figs. 12–15 for the four studies [23–26], respectively. The model captured well the uptake and clearance of nicotine from cigarette, with model predictions generally being within one standard deviation of the data. For these simulations, we treated each repeated use, the same as the first use. That is, the mass of nicotine inhaled with each puff within a simulation was identical, regardless of puff volume, and only the estimated mass taken is varied across the study. This assumption held up for each of the two studies with repeated use in Fig. 14 [25] and Fig. 15 [26], where the model did an excellent job at capturing the overall pattern of nicotine in plasma even following repeated smoking of conventional cigarettes.
ENDS products
The next example of the model application was focused on evaluating uptake of nicotine following use of ENDS products. For evaluating our PBPK model’s ability to predict the time course of nicotine following use of ENDS products, we simulated and compared our PK curve predictions against measured nicotine PK profiles from a study conducted by Lopez and coworkers[28]. In that study, the researchers measured and reported data over two consecutive uses of an ENDS product (eGo device; 3.3V, 1000 mAh battery with a 1.5 Ω, dual-coil, 510-style cartomizer). Each of the 16 ENDS-naïve cigarette smoker study participants who completed the study (defined in their publication as healthy, aged 18–55 years, used at least 15 cigarettes daily, and used an e-cigarette less than 5 instances in their lifetime) participated in four randomized sessions. Depending on the randomization sequence, a participant was provided with a product containing e-liquid with either 0, 8, 18, or 36 mg/mL of nicotine, in each of their four sessions. In each session, the study participants, were asked to take a total of 10 puffs with a 30 s interval per use (1h interval between the two uses). No data were provided on the amount of nicotine intake from the device.
Since the amount of nicotine inhaled per puff for the three nicotine strengths is not specified in the publication [28], we needed to calculate those from other information provided in the publication. The amount of aerosol generated per puff depends on the power input and puff duration. Although this can vary among different devices, we follow a semi-empirical approach. In general, of the total amount of electrical energy supplied to the device, a portion is used for heating the system. The rest is used to heat and vaporize the e-liquid, which is generally a mixture of propylene glycol, glycerine, water, and nicotine. Using thermodynamics values, the total amount of energy (sensible and latent) needed to vaporize 1 mg of an e- liquid mixture is approximately 1.5 J. This is the estimated energy needed to heat the e-liquid from the room temperature to approximately 275°C. Lopez et al. [28] provided the battery voltage (3.3 V) and the heater resistance (1.5 Ω), and reported puff durations ranging from 2.2 to 2.9 s. The power input is calculated to be 7.26 W, and the total energy inputs were 21, 20.38, and 16 J for the 8, 18, and 36 mg/mL e-liquids, respectively. Out of this energy, 8.7 J is used to heat up the system, which corresponds to the energy consumed before any aerosol is formed [29]. Subtracting this energy from total energy and dividing by 1.5 (J/mg), we estimated the aerosol mass per puff, and the corresponding nicotine inhaled per puff as 0.066, 0.14, and 0.176 mg/puff for the 8, 18, and 36 mg/mL nicotine levels, respectively. Assuming 90% of inhaled aerosol is absorbed with 10% exhaled, the final nicotine dose inputs to the PBPK model used were 0.06, 0.126 and 0.158 mg/puff of nicotine, in the order of increasing nicotine level in the e-liquid.
Following the same procedure introduced for the dosimetry model, the distribution of absorption in the RT of 10% in BC, 15% in URT and 75% in LRT, was used for modeling this scenario. For this distribution, we also had to calculate the vapor-particle partitioning of nicotine. Since no information on the pH of the e-liquid was given in the paper, an activity coefficient of 10 was used to correct the Raoult’s law for nicotine partitioning between liquid and vapor phase over liquid mixture. Such a large activity coefficient is not unusual for low molar fraction of nicotine in the liquid mixture. No transfer to GI tract was assumed, because only 10% of nicotine is absorbed orally and the use duration is only 5 min per session.
Next, we employed the permeation model to estimate deposition location specific transfer rates of nicotine to the circulation system (similar to Fig. 7a–c, but different ratios) for each region of the airway; BC, URT, and LRT, respectively. They are presented separately as these transfers occur on significantly different timescales in these three regions. The blood flow rates vary in different regions and as previously discussed they are treated differently in the PBPK model.
A comparison of the predicted venous nicotine plasma time-course to the experimental data is presented in Fig. 16. The error bars in the figure represent one standard deviation. In this case, however, the assumption of identical intake across two uses (60 min apart) was not as fruitful as it was with the simulations for cigarette smoking. Our PBPK model provided excellent agreement with the first use, but slightly over predicted the time-course plasma for higher nicotine concentration e-liquids, upon the second use. The mean concentrations from the model predictions all lie within one standard deviation of the reported data. Given that the uptake also depends on the depth of inhalation, which is not specified in the study, a possible explanation may be that subjects altered their intake rate by reducing their depth of inhalation for e-liquid with high nicotine concentration. Lower depth of inhalation results in lower absorption for this case and lower measured nicotine concentration in plasma. Based on sensitivity analysis that we performed with our model, depth of inhalation would have a significant impact on the overall outcome of the simulation. In general, a shallow depth of inhalation results in less aerosol reaching the LRT, thus less uptake and more exhale. In addition, the subjects were described as naïve users of ENDS products, who were predominantly smokers of conventional cigarettes, and this could also have played a role in the altered intake with the second use. Other explanations related to the topography changes that may affect the actual PK profile are also feasible and require further investigation.
Vapor inhaler
The next example is related to a vapor inhaler. For this, we simulated the study of Liu and coworkers [30] and compared our predicted results with the outcomes reported in their clinical study. In their study [30], nicotine venous plasma concentrations were measured, during and after use of the Nicotrol® nicotine inhaler. Individual subjects were asked to take 80 puffs (2 s puff every 15 s) over a 20 min period. Based on the manufacturer’s information [31], the total mass inhaled over 80 puffs was 2 mg. From the vapor absorption-based dosimetry model described earlier, it was estimated that 95% of nicotine is absorbed in the BC and 5% in the URT airway; 70% of the orally absorbed nicotine was assumed to transfer to GI tract, as the 20 min use duration included multiple instances of swallowing. The rate of nicotine transfer to the plasma for vapor inhaler was estimated using the permeation model, and is presented in Fig. 8.
Figure 17 compares the simulated PK profile following use of a vapor inhaler and compares the model predictions to experimental data [32] PK profiles. Overall, the PBPK model well captured the slow and muted rise to maximum concentration which occurred near the end of exposure (i.e., 20 min), which was driven by a larger proportion of the inhaled mass being transferred to the GI tract following oral absorption. Good agreement between the model predictions and mean of the experimental data further affirms the validity of applying our PBPK to predict nicotine PK from another type of nicotine containing product.
Smokeless tobacco
Among the different product uses, this was the most challenging product use to simulate. The mass of nicotine in the product is extracted with saliva in the BC, where it is either absorbed locally, swallowed to the GI tract, or removed via spittle through expectoration. To evaluate the PBPK model’s viability to predict nicotine PK in users of oral tobacco products, we simulated the study of Digard et al. [33], which reported the venous plasma time-course of nicotine during and after the use of different smokeless tobacco formats; subjects were asked to keep the loose snus (10.79 or 27.09 mg nicotine/plug) or pouched snus (10.72 or 14.67 mg nicotine/pouch), in place for the first hour of the study. The authors also reported the nicotine mass remaining in the products after use, which allowed the estimation of the nicotine extracted over the use period. The amount of nicotine available for absorption by the body is the difference between the pre-use and post-use nicotine content. However, for the PBPK model, in addition to the amount extracted, the time profile of nicotine release over the 60 min use period is needed. Following the procedure described for smokeless tobacco (Figs. 4 and 5), the rate of nicotine release in the BC was approximated as exponentially decreasing over time, m = m0 e-bt, where m0 is the rate of nicotine release at time 0 (mg/min), b is the rate constant (1/min) and t is the time in minutes. Values of m0 and b are selected such that the total mass released during the usage period matched the measure values reported by Digard et al. [33], and are shown in Table 3. For smokeless tobacco, 100% of the nicotine absorption is assumed to take place in the BC, with no swallowing. The relative transfer rate of nicotine from the BC to the plasma, corresponding to nicotine release/dissolution presented in Fig. 6, was estimated using the previously described permeation model. The predictions for relative transfer rate at which nicotine is removed from the BC to the plasma is shown in Fig. 9.
Table 3
Constants for the rate of nicotine release in the BC during the 60 min use time matching data of Digard et al. [33]: m = m0 e-bt
Product
|
m0 (mg/min)
|
b (1/s)
|
Loose snus (10.79 mg/mL)
|
0.0678
|
0.006
|
Loose snus (27.09 mg/mL)
|
0.120
|
0.004
|
Pouched snus (10.72 mg/mL)
|
0.065
|
0.006
|
Pouched snus (14.67 mg/mL)
|
0.0879
|
0.006
|
m0 rate of nicotine release at time 0, b rate constant, t time. |
As with the previous simulations, the comparison of model predicted and measured nicotine plasma concentrations shown in Fig. 18, demonstrate that predictions were within one standard deviation of reported data. The mean predicted nicotine levels were all within the individual variation of PK data for study participants in the study [33], further affirming the our PBPK model’s prediction capabilities.