Understanding Pulmonary and Systemic Pharmacokinetics of Inhaled Fluticasone Propionate in Healthy Subjects Using Population PK Analysis

Abstract

Background and Objectives

One of the main hurdles in pulmonary drug delivery is the inability to directly quantify drug concentrations at the site of action within the lungs, which are critical for therapeutic efficacy. This research attempted to address this limitation by applying population pharmacokinetic modeling to characterize the pharmacokinetics of fluticasone propionate (FP), integrating multiple pulmonary absorption pathways and identifying key covariates that influence its pharmacokinetics.

Methods

Plasma concentration-time data from 96 healthy volunteers participating in four pharmacokinetic (PK) studies using two FP formulations were analyzed, implementing non-linear mixed effects modeling to estimate population pharmacokinetic parameters, including lung deposition-related parameters. In all studies, 1000-µg doses of two FP formulations were administered via a metered-dose inhaler (MDI). A nonlinear mixed effects analysis with a maximum likelihood expectation maximization approach was applied using ADAPT 5 software.

Results

A three-compartment model with first-order distribution, elimination, and absorption processes described the FP pharmacokinetics. The final model considered three absorption peaks, in order of appearance in plasma, presumably in the central and peripheral parts of the lungs, with a delayed peak arising from outside the lung compartments. Weight was a predictor of the apparent clearance and volumes of distribution of FP. The final model suggested that approximately 17% of systemic exposure was associated with drug absorption from the central lung regions, while approximately 14% was associated with absorption from the peripheral lung regions. A secondary delayed peak was observed in the population PK analysis, with the remaining proportion of systemic exposure (~69%) modeled as arising outside the lung compartments. The precise origin of this delayed component remains uncertain, particularly given the extensive first-pass metabolism reported for fluticasone, and may reflect multiple overlapping processes, including redistribution or secondary pulmonary absorption.

Conclusion

The proposed population PK model was able to characterize the PK of FP administered via inhalation and may be useful for better understanding the PK of other drug products administered through the inhalation route.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s13318-026-00989-0.

Key Points

The PopPK model successfully captured inter-individual variability in fluticasone propionate kinetics using a three-compartment structure with dual lung absorption pathways

Weight emerged as a significant covariate, influencing both clearance and distribution volumes, underscoring the importance of body size in dose optimization

This modeling framework offers a scalable approach to evaluate inhaled drug absorption and may inform future PopPK studies of respiratory therapeutics

Introduction

Determining the fraction of an inhaled drug that reaches the lungs is critical for optimizing the efficacy and safety of inhaled therapies. However, this area has significant challenges. The complex anatomy of the respiratory tract, physiological and formulation factors, variations in individual breathing patterns, and differences in disease states can all influence lung deposition. Additionally, the choice of the delivery device (e.g., nebulizers, dry powder inhalers, or metered dose inhalers) and the inhalation technique introduce variability, which further complicates assessment of lung deposition. Compounding these issues is the limited availability of non-invasive tools to quantify regional lung deposition in real time. Imaging techniques such as a gamma scintigraphy or PET/CT offer valuable insights but are resource-intensive and not widely accessible [1]. The inability to measure local lung concentrations responsible for lung efficacy is a common challenge in any inhalation drug delivery program targeting the lungs [2]. There is therefore a need to build improved models that can predict the fate of therapeutic agents administered to humans via the pulmonary route with the desired precision and reproducibility.

Population pharmacokinetics (PopPK) enables the quantitative assessment of drug concentration-time profiles across individuals within a population, while accounting for variability due to physiological, demographic, and environmental factors. By modeling both fixed (typical PK parameters) and random effects (interindividual variability and residual unexplained variability), PopPK can provide insights into drug deposition patterns not apparent in noncompartmental or individual PK analyses.

This research presents the development and validation of a population pharmacokinetic (PopPK) model for fluticasone propionate (FP), a synthetic, trifluorinated glucocorticoid, using rich Phase 1 data from healthy subjects. The population PK model attempted to distinguish between pulmonary and gastrointestinal absorption contributions and identify covariates that influence systemic exposure.

This study contributes novel insights by explicitly modeling multiple pulmonary absorption pathways, which better reflect the heterogeneous deposition and clearance mechanisms of inhaled particles and demonstrates the utility of an absorption model even in healthy subjects. Distinguishing between absorption pathways enhances physiological plausibility and allows extrapolation to disease settings in which lung morphology may alter deposition. A flexible modeling tool (ADAPT 5) was used, which enabled custom model coding. A robust four-level validation provided strong evidence of accurate individual fits, as well as PK model reliability and generalizability within the studied population.

The ability to simulate systemic exposure using different absorption processes for an inhaled drug may be useful to inform dose selection in vulnerable populations (e.g., pediatrics, the elderly), guide bioequivalence assessment for generic inhalers, and support model-informed drug development (MIDD) submissions to regulatory agencies. This research, therefore, contributes toward building the foundation for predicting variability in systemic exposure of inhaled drugs due to formulation or device changes and for developing future population PK/pharmacodynamic (PK/PD) models.

Materials and Methods

Study Design

The dataset was prepared by combining all available individual PK data from four studies, including two formulations of fluticasone propionate whose PKs were assessed in 96 healthy subjects (3840 observation data points) and incorporated covariate information. Details of the four PK studies included in the current analysis are summarized in Table 1. All trials received prior approval from the Dakshata Independent Ethics Committee, Mumbai, India (08-11-261, 10-04-027, 10-11-066, and 11-08-041), confirming ethical compliance before initiation, and were conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. Written informed consent was obtained from all subjects prior to study entry.

Table 1.

Studies included in the population PK analysis studies of fluticasone propionate

Study number	Study design	Subjects (n)	Treatment dose (single dose)	Number of PK sampling timepoints post dosing
Study 1	Randomised, crossover study	24	Formulation 1: 1000 µg Formulation 2: 1000 µg	20
Study 2	Randomised, crossover study	22	Formulation 1: 1000 µg Formulation 2: 1000 µg	20
Study 3	Randomised, crossover study	22	Formulation 1: 1000 µg Formulation 2: 1000 µg	20
Study 4	Randomised, crossover study	28	Formulation 1: 1000 µg Formulation 2: 1000 µg	20

All studies were open-label, single-center, randomized, crossover trials that assessed the safety, tolerability, and PK following single doses of fluticasone propionate. Each subject received a single dose of FP, 1000 µg, delivered as four consecutive 250 µg inhalations via a metered-dose inhaler within a 2-min window. No spacer or aerochamber device was used during drug administration. Healthy male subjects aged 18–45 years inclusive with a body weight > 45 kg and a body mass index (BMI) between 18.5 and 30.0 kg/m² were included. In all studies, healthy subjects were exposed to two FP HFA formulations delivered via a metered-dose inhaler in a crossover manner under fasting conditions with a 7–14-day washout period. Both formulations had a comparable fine particle mass (FPM), i.e., the fraction or percentage of the drug mass contained in an aerosol cloud that may be small enough to enter the lungs and exert a clinical effect, and primarily differed in the source of the active pharmaceutical ingredient.

PK Assessments

Serial blood samples for PK analysis were collected before dosing and at various time intervals following administration of the single doses of FP. Blood samples for PK analysis were collected in vacutainer tubes containing K2EDTA. Plasma was separated from each blood sample by centrifugation and immediately frozen at ≤ − 20 °C until analysis. For all studies included in the population PK analysis, the plasma FP concentrations were determined using the same validated liquid chromatography with tandem mass spectrometry (LC-MS/MS) method with a lower limit of quantification (LLOQ) of 9.9 pg/ml. The linearity range for FP was from 9.9 to 742.4 pg/ml.

Population Pharmacokinetic Model Development

A multi-step pharmacometric analysis was employed to characterize the population pharmacokinetics of inhaled FP. The analysis included exploratory data analysis, population pharmacokinetic (PopPK) model development, discrimination, validation, and simulations. All population pharmacokinetic analyses were conducted using ADAPT 5^® (version 65) with the MLEM algorithm and 3000 importance sampling points, assuming a ln-normal distribution of PK parameters [5]. Convergence was deemed to be attained with any single model when all PK parameters were fitted visually to their final value. Usually, this was attained by population iteration 800 or before. Then, the median value of the objective function was calculated over the next 200 consecutive population iterations (e.g., between 800 and 1000), and the iteration at convergence (reported in this manuscript as “OFV”) was set to be the first population iteration reaching that exact median objective function value [6].

All available FP concentration data points were fitted to population PK models of increasing complexity. These included one to four compartment models, with one to four absorption peaks and linear distribution and elimination characteristics. The final most probable model was selected according to the parsimony principle, which was the one that best fit the data according to the MLEM objective function (OFV), assuming that the latter followed a chi-square distribution. A model with one additional parameter was considered significantly better if it reduced the MLEM objective function by at least 3.84, corresponding to a critical p-value below 0.05. The final most probable population PK model was parameterized using three separate absorption peaks with their associated lag times, a total plasma clearance (CL/F), a central volume of distribution (Vc/F), distributional clearances between the central compartment and either the first (CLD/F) or second peripheral compartment (CLD2/F), and the volume of distribution for either the first (Vp/F) or second peripheral compartment (Vp2/F). The residual variability was fitted using a combined additive proportional error model. For detailed information on the model code, refer to the supplementary data.

A schematic of the final population PK model is shown in Fig. 1.

Fig. 1.

Final population model describing the PK of fluticasone propionate. kaINHD1c Absorption rate constant in the central lung compartment for formulation 1, kaINHD1p absorption rate constant in the peripheral lung compartment for formulation 1, kaINHD2c absorption rate constant in the central lung compartment for formulation 2, kaINHD2p absorption rate constant in the peripheral lung compartment for formulation 2, kaINHGD1 absorption rate constant in the gut for formulation 1, kaINHGD2 absorption rate constant in the gut for formulation 2, TlagD1c time lag in the absorption in the central lung for formulation 1, TlagD1p time lag in the absorption in the peripheral lung for formulation 1, TlagD2c time lag in the absorption in the central lung for formulation 2, TlagD2p time lag in the absorption in the peripheral lung for formulation 2, TlagGD1 time lag in the absorption in the gut for formulation 1, TlagGD2 time lag in the absorption in the gut for formulation 2, CL/F plasma clearance (central compartment), CLD/F plasma clearance (peripheral compartment 1), CLD2/F plasma clearance (peripheral compartment 2), Vc/F central volume of distribution, V_p/F and Vp2/F = peripheral volume of distribution. *Average body weight (ABW) added as a covariate

Covariate Modeling

The potential relationships of covariates (age, height, BMI, FEV₁% predicted at baseline, and weight) and pharmacokinetic parameters were explored. The covariates were tested for their impact on the PK parameters, with a reduction in the objective function value (OFV) according to a chi-square distribution and a p-value of 0.05 serving as criteria for model selection. Only body weight showed correlation with the PK parameters; hence, it was investigated in the final model.

Internal Validation

The following four steps were followed to validate the final population PK model [6].

• Verification of the appropriateness of goodness-of-fit (GOF) plots for individual fitted vs. observed PK data.

• Verification that the individual fitted parameters agree with the observed noncompartmental PK results.

• Verification of the GOF plots for population vs. observed data and verification that the population predicted parameters agree with the individual fitted PK parameters.

• Verification that visual predictive checks (VPCs) are appropriate.

Results

Demographic Data

The population PK analysis used all concentration versus time data from 96 subjects who received both formulations. The summary statistics of baseline subject descriptors for the analysis population are presented in Table 2. All participants were male and non-smokers.

Table 2.

Demographic characteristics of subjects (N = 96) across the four studies

Variable (units)	Value/status Mean (SD/percentage)
Age (years)	27 (5.34)
Sex	Male (100 %)
Height (meters)	1.68 (0.05)
Weight (kg)	62.45 (7.42)
BMI (kg/m2)	21.90 (1.96)
FEV1 (% predicted)	91.89 (8.17%)

Model Development

Population PK Model

Several structural and residual error models, which included one-, two-, and three-compartment models, were tested and compared to determine which one best described the plasma concentration-time profile of FP. Among the one-, two-, and three-compartment models, the three-compartment model had the lowest statistically significant objective function value and best fit the data; hence, it was considered the final model. The final model included estimating the percentage systemic exposure presumably from the central and peripheral compartments of the lung. Absorption lag times were included for the central and peripheral lung compartments and the gut. Table 3 presents the results associated with the main models that were compared with each other to find the final PopPK model.

Table 3.

Details of the main models tested

Name of the model	Number of parameters	Run	Ref Run	OFV	Δ OFV (p < 0.05)
1 CPT 1 KA	7	1		Could not converge	N/A
2 CPT 1 KA	9	2		29555.5	N/A
2 CPT 2 KA	15	3	2	27825.4	− 1730 (≤ − 12.592)
1 CPT 2 KA	13	4	3	28404.8	599.4 (≥ 5.991)
2 CPT 1KA 1K0	15	5	4	29231.5	826.7 (≥ 5.991)
3 CPT 2 KA	17	6	5	27765.4	− 1466.1 (≤ − 5.991)
Final structural model* 3 CPT 3 KA	23	7	6	27630	− 135.4 (≤ − 12.592)
2 CPT 4KA	27	8	7	28113.5	483.5 (≥ 9.488)
3 CPT 4 KA	29	9	7	28142.1	512.1 (≥ 12.592
Final model with covariate** 3CPT3KA with ABW on clearances and volume of distributions	23	10	7	27522.7	− 107.3 (≤ − 3.841)

ABW average body weight, cpt compartment, KA absorption rate constant (first order), K0 absorption rate constant (zero order), OFV objective function value, Δ OFV change in objective function value, Ref reference

*Final structural model, **Final covariate model

PopPK Model Covariate Testing

Significant covariates included total body weight on the apparent volume of distribution and apparent clearance (significant reduction in − 107.3 objective function points), both with their expected allometric exponents of 1 and 0.75, respectively [7] (Table 3).

The final model suggested that approximately 17% of systemic exposure was associated with drug absorption from the central lung regions, while about 14% of the systemic exposure was associated with absorption from the peripheral lung regions. A secondary delayed peak was observed in the population PK analysis, with the remaining proportion of systemic exposure (~ 69%) modeled to arise outside the lung compartments. The precise origin of this delayed component remains uncertain, particularly given the extensive first-pass metabolism reported for fluticasone, and may reflect multiple overlapping processes, including redistribution or secondary pulmonary absorption (Tables 4 and 5). The residual variability in fitted concentrations of fluticasone propionate was 8.97%.

Table 4.

Population predicted and individual fitted PK absorption parameter estimates—final PopPK model

PK parameters	Units	Individual fitted		Population predicted
		Geomean	GeoCV (%)	Geomean	GeoCV (%)
Absorption parameters—Formulation 1
kaINHD1c	h−1	1.16	45.3	1.15	46.8
kaINHD1p	h−1	0.82	63.1	0.77	51.2
kaINHGD1	h−1	0.07	8.9	0.07	9.6
% Systemic exposure due to central lung absorption	%	14.3	53.1	13.4	55.4
% Systemic exposure due to peripheral lung absorption	%	13.4	56.4	11.9	57.8
TlagD1c	h	0.04	30.4	0.04	31.6
TlagD1p	h	0.17	15.2	0.17	20.4
TlagGD1	h	0.61	52.4	0.60	47.5
Absorption parameters—Formulation 2
kaINHD2c	h−1	0.82	47.6	0.76	44.7
kaINHD2p	h−1	0.65	63.7	0.59	56.8
kaINHGD2	h−1	0.07	18.7	0.07	17.9
% Systemic exposure due to central lung absorption	%	20.0	33.9	20.0	34.5
% Systemic exposure due to peripheral lung absorption	%	15.3	31.6	14.7	32.1
TlagD2c	h	0.04	23.8	0.04	24.2
TlagD2p	h	0.14	47.9	0.14	50.7
TlagGD2	h	0.37	63.5	0.37	63.4
Frel	–	1.05	47.2	1.02	34.2

Geomean geometric mean, GeoCV geometric coefficient of variation, h hour, kaINHD1c absorption rate constant in central lung cpt, kaINHD1p absorption rate constant in peripheral lung compartment for formulation 1, kaINHD2c absorption rate constant in the central lung compartment for formulation 2, kaINHD2p absorption rate constant in the peripheral lung compartment for formulation 2, kaINHGD1 absorption rate constant in the gut for formulation 1, kaINHGD2 absorption rate constant in the gut for formulation 2, TlagD1c time lag in the absorption in the central lung for formulation 1, TlagD1p time lag in the absorption in the peripheral lung for formulation 1, TlagD2c time lag in the absorption in the central lung for formulation 2, TlagD2p time lag in the absorption in the peripheral lung for formulation 2, TlagGD1 time lag in the absorption in the gut for formulation 1, TlagGD2 time lag in the absorption in the gut for formulation 2

Table 5.

Population predicted and individual fitted PK systemic parameter estimates—final PopPK model

PK parameters	Units	Individual fitted		Population predicted
		Geomean	GeoCV (%)	Geomean	GeoCV (%)
Systemic parameters—Formulation 1/2
Vc/F (*)	L	118.7	77.4	111.0	70.3
Vp/F (*)	L	83.9	92.6	80.3	84.4
Vp2/F (*)	L	1089.1	44.8	1070.0	46.2
CL/F (*)	L/h	146.3	25.9	145.0	27.0
CLD/F (*)	L/h	263.4	62.7	248.0	60.2
CLD2/F (*)	L/h	2164.7	63.2	2030.0	61.9

Literature evidence: Krishnaswami et al., estimated CL/F to be 46 l/h (calculated assuming a systemic bioavailability of 10%) post dosing with 500 µg FP via DPI [8]; Drescher et al., administered 500 µg FP via DPI wherein estimated CL/F was fixed to 72.2 l/h [9]; Mackie et al., administered FP IV 250–1000 µg wherein CL ~ 66 L/h, V_ss ~ 318 (CL was dose-independent) [10].

*Geomean value for the median 62 kg individual

Geomean geometric mean, GeoCV geometric coefficient of variation, CL/F plasma clearance (central compartment), CLD/F plasma clearance (peripheral compartment 1), CLD2/F plasma clearance (peripheral compartment 2), V_c/F central volume of distribution, V_p/F and Vp2/F peripheral volume of distribution

PopPK Model Validation

Step 1: Goodness-of-Fit Plots for Individual Fitted vs. Observed

Goodness-of-fit (GOF) plots, including individual fitted PK values versus observed PK values, were created and closely inspected to ensure there were no obvious trends or biases in the fitted data. Scatter plots of observed versus individual fitted concentrations indicate that the model structure is good. This is presented in Fig. 2.

Fig. 2.

Goodness-of-fit plots for individual fitted vs. observed data—final PopPK model

Step 2: Fitted vs. Observed PK Parameters

The exposure parameters C_max, AUC_0−t, and T_max were calculated via noncompartmental analysis using the individual fitted concentrations from the model, and they were compared to values obtained using individual observed data. This was done to ensure that the model correctly characterized the observed data at the individual level. Results are presented in Table 6. Mean model-fitted and observed PK parameters were very similar. The fitted concentrations indicated that the final popPK model can adequately reproduce the results from the noncompartmental analyses using observed data. This confirms the suitability of the final PopPK model's structure for individual profiles.

Table 6.

Comparison of noncompartmental PK parameters calculated using observed and model-fitted data

PK parameter		Formulation 1	Formulation 2
		Geomean (%CV) (pg/ml)	Geomean (%CV) (pg/ml)
Cmax	Observed	140.28 (58.50)	155.69 (52.96)
	Fitted	131.34 (53.54)	146.78 (45.98)
	Ratio (fit/obs)	0.93	0.94

		Geomean (%CV) (pg*h/ml)	Geomean (%CV) (pg*h/ml)
AUC0−t	Observed	1429.94 (56.81)	1393.49 (56.06)
	Fitted	1451.32 (57.35)	1414.21 (45.46)
	Ratio (fit/obs)	1.01	1.01

		Median (min–max) (hours)	Median (min–max) (hours)
Tmax	Observed	1.84 (0.5–5)	1.58 (0.5–4)
	Fitted	1.50 (0.5–4)	1.50 (0.5–4)

AUC_0−t area under the curve from time 0 to the last measurable time point (t), C_max maximum concentration, CV coefficient of variation, Geomean geometric mean, pg picograms, Obs observed, T_max time of maximum concentration

Step 3: Final Model—Population Predictions vs. Observed Data

The GOF plots of population predicted vs. observed data indicated no obvious trends or biases in the predictions (Fig. 3a). Weighted residual plots, displaying the differences between observed and population predicted values, exhibited a random distribution around zero (Fig. 3b). The dotted line in the figures represents the identity line. As shown in Tables 4 and 5, the population predicted parameter estimates were suitably similar to the individual fitted values; this was further validated in step 2. The final PopPK model population predicted results were therefore appropriate and could be used to simulate other studies/data, pending successful completion of the last validation step.

Fig. 3.

a Goodness-of-fit plots for population predicted vs. observed data—final model. b Weighted residuals vs model prediction—final model

Step 4: Population PK Visual Predictive Checks (VPCs)

The final model was used to simulate concentration-time profiles for completely virtual subjects, similar in their demographic characteristics to those included in the PopPK analysis. Using these simulated PK values, median and 90% confidence intervals were constructed (using a percentile approach) and overlaid on the observed data for each product.

Figure 4a and b illustrates the visual predictive checks for formulation 1 and 2.

Fig. 4.

a Visual predictive check for formulation 1 of fluticasone propionate. b Visual predictive check for formulation 2 of fluticasone propionate

The visual predictive check (VPC) results demonstrate that the population pharmacokinetic model adequately describes the observed data. The close alignment of the observed median values with the simulated median line suggests that the model's central tendency is accurate, while the coverage of the 5th and 95th percentiles indicates that variability is well captured. The VPCs indicate that the model is appropriate for predicting new populations of subjects and is therefore suitable for simulation-based inference.

Discussion

Constructing PK models that can closely characterize systemic exposure from the lungs is challenging for many reasons. The human lung microenvironment is complex and further complicated for inhaled drugs by device designs, aerodynamic particle sizes, distribution of the dispersed inhaled product, fluid mechanics, physiological factors in the lungs, and the way the patient interacts with the drug delivery device [2]. In this research, a population-pharmacokinetic modeling approach was used to investigate the pulmonary and systemic exposure of FP in healthy subjects. Pharmacokinetic data from four studies were used to describe the PK of fluticasone propionate following administration of two FP formulations as a “single dose” of 1000 µg in healthy subjects. Several structural and residual error models, which included 1-, 2-, and 3-compartment models, were tested to find the one that best described the plasma concentration-time profile of FP. The three-compartment model had the lowest objective function value and best fit to the data; hence, it was considered the final model. Bodyweight was a significant covariate in the apparent volume of distribution and apparent clearance of fluticasone based on covariate analysis.

As seen in the GOF figures for the individual fitted versus observed data and population prediction versus observed data, no obvious bias or abnormal trends were apparent. Furthermore, the population systemic parameter values are suitably similar to the individual parameter values. The visual predictive checks confirmed that the model is appropriate for simulating new subjects.

Drug clearance of drug particles from the lung depends on whether particles are deposited in the central or the peripheral part of the lung, as mucociliary clearance in the conducting airways is faster than the slow macrophage clearance in the alveolar space [11]. In our study, the population PK model, which characterized the pharmacokinetics of FP after inhalation, also provided estimates of the contribution to the systemic exposure presumably from the central and peripheral lung compartments. The model delineated contributions to systemic exposure from distinct lung regions, estimating ~ 17% from the central lung and ~ 14% from the peripheral lung.

A secondary delayed peak was also observed, accounting for ~ 69% of systemic exposure, and was associated with a significant lag time of > 30 min and was therefore presumed to be gut absorption. While this component was modeled as arising outside the lung compartments, its precise origin remains uncertain. Given FP’s extensive first-pass metabolism and limited gastrointestinal absorption, the delayed peak may reflect overlapping processes, including redistribution or secondary pulmonary absorption; hence, the apparent contribution from the gut is likely to be overestimated.

Fluticasone propionate (FP) has been reported to have low oral systemic bioavailability, primarily and presumably due to limited gastrointestinal absorption and extensive first-pass metabolism in the liver [12]. Some studies, however, have reported oral systemic availability of around 16% [12]. The absence of an intravenous comparator in our dataset limits the ability to draw firm conclusions regarding oral contributions to systemic exposure. The estimates rely on assumptions regarding absorption pathways, particularly that the delayed peak reflects gastrointestinal uptake rather than a third absorption peak from the lung. As such, these observations are hypothesis-generating and not definitive. Future studies incorporating intravenous administration or charcoal co-treatment are required to validate these exploratory findings.

The absorption lag times estimated in our model (~ 2–3 min for the central lung and ~ 9–10 min for the peripheral lung) align with mechanistic considerations described by Hastedt et al. [13], who highlighted the influence of drug lipophilicity, particle size, solubility in airway lining fluids, and regional physiology on dissolution and absorption. These results underscore the importance of regional deposition in shaping systemic exposure and highlight the need for physiologically relevant models that capture device- and patient-specific factors.

The development of population pharmacokinetic (PopPK) models for fluticasone propionate (FP) has mirrored the growing sophistication of regulatory expectations for inhaled drug products. Early models, such as that by Derendorf et al. [14], introduced dual-absorption frameworks to distinguish between pulmonary and gastrointestinal uptake, laying the foundation for inhaled drug modeling. However, most subsequent studies adopted simpler one- or two-compartment structures, often with first-order absorption kinetics and limited resolution of regional lung deposition. These models were largely based on dry powder inhaler (DPI) data. While useful for estimating systemic exposure, they lacked the granularity needed to simulate device-specific deposition or absorption in diseased airways. Notably, Mehta et al. and Xu et al. were among the few to explore metered-dose inhaler (MDI) formulations of fluticasone, though their structural simplicity limited mechanistic insight [15, 16].

As regulatory agencies such as the FDA and EMA began to emphasize model-informed drug development (MIDD), the need for more physiologically relevant models became clear. The FDA’s OINDP guidance [17] and EMA’s evolving inhalation product frameworks [18] encouraged developers to incorporate device performance, regional deposition, and IVIVC into their modeling strategies. This shift culminated in more recent efforts, such as those of Drescher et al. [9], which applied refined compartmental models to asthma populations using FP DPI [10].

This research further advances this trajectory by introducing a three-compartment PopPK model tailored to FP delivered via MDI. Beyond characterizing FP disposition in healthy subjects, this model introduces region-specific absorption lag times that provide mechanistic insight into pulmonary drug uptake. Such granularity has not been widely reported in prior FP models, which often relied on simpler compartmental structures. By aligning with evolving regulatory priorities for model-informed drug development, this framework has potential utility in bioequivalence assessments, device optimization, and extrapolation to patient populations with altered airway physiology. While the delayed secondary peak remains exploratory, its identification underscores the need for confirmatory studies and highlights the complexity of pulmonary absorption processes. Thus, this work contributes both mechanistic and translational values to the field of inhaled drug development.

Further investigations comprising in vitro experiments, additional pharmacokinetic studies, and more sophisticated mathematical approaches could be necessary to elucidate the highly complex pulmonary processes and aid in developing a more robust in vitro-in vivo correlation for the inhaled drug fluticasone propionate.

Gender effect was not evaluated in the model as only healthy male subjects were included. The studies were conducted in healthy subjects, which may limit generalizability to patients with asthma or COPD who have altered airway structures, inflammation, and mucociliary clearance [19]. Though not discussed in this article, the doses tested were safe and well tolerated in healthy subjects.

Conclusions

The PopPK model provides a physiologically consistent characterization of FP disposition following inhalation. The PK of fluticasone propionate was adequately described by a linear three-compartment model with first-order absorption and absorption lag times. While the findings regarding delayed systemic exposure are exploratory, they highlight the complexity of pulmonary drug absorption and the need for confirmatory studies. This work contributes to the growing body of evidence supporting model-informed approaches for inhaled drug development and underscores the importance of integrating device performance, regional deposition, and patient-specific factors into PK modeling strategies.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open access funding provided by Symbiosis International (Deemed University). This research received no external funding.

Declarations

Conflict of Interest

The authors declare no conflict of interest.

Ethical Approval

The study was conducted in accordance with the Declaration of Helsinki. The Independent Institutional Review Board approved all four PK studies included in the datasets used for the PopPK model. The EC approval code (EC approval dates in dd/mm/yyyy format) of the four PK studies are as follows: 08-11-261 (14/12/2008), 10-04-027 (12/09/2010), 10-11-066 (18/01/2011), and 11-08-041 (10/11/2011).

Consent to Participate

All participants provided written informed consent prior to trial-related procedures.

Availability of Data and Material

Restrictions apply to the availability of these data. Data were obtained from Cipla Ltd. and will require permission from Cipla Ltd. before the raw data are shared.

Code Availability

Final model code is provided as Supplementary1_WtClmodel.for.

Author Contributions

Conceptualization: J.R. and S.S. Methodology, formal analysis: J.R. and M.D., writing-original draft preparation: J.R., review and editing-S.S. and M.D., supervision: M.D. and S.S., study conduct: N.M., bioanalysis: R.N. and J.P. All authors have read and agreed to the published version of the manuscript.

Consent to Publish

All study participants provided informed consent for the publication of anonymized data and findings. No identifying information is included in the manuscript.