Pharmacokinetics of repeated, high-dose fentanyl administration in rats

Abstract

Fentanyl is a potent synthetic opioid widely used as a recreational drug. It is primarily metabolized in the liver to the inactive compound norfentanyl, with only a small fraction excreted unchanged via the kidneys. However, it remains unclear whether repeated high-dose exposure alters its pharmacokinetics (PK) over time. This study evaluated the effects of repeated high-dose fentanyl administration on PK and assessed sex-related differences in metabolism.Male and female Sprague–Dawley rats (n = 19) received intravenous boluses of 100 μg/kg fentanyl twice weekly for five weeks. After the 1st, 5th, and 10th doses plasma concentrations of fentanyl and norfentanyl were measured using ultra-high-performance liquid chromatography - electrospray ionization - tandem mass spectrometry. PK data were analyzed using non-compartmental methods, followed by covariate analysis within a population PK model and a physiologically based PK model to identify underlying physiological factors.Non-compartmental analysis showed significantly lower C_max and longer t_1/2 of norfentanyl in females compared to males. Population PK analysis using a two-compartment model confirmed slower fentanyl clearance and reduced norfentanyl formation in females. Repeated dosing led to decreased fentanyl clearance after the 10th exposure. In males, higher muscle volume was inversely correlated with total body clearance, while in females, fat volume correlated positively with total fentanyl exposure.Repeated high-dose fentanyl administration had only modest effects on PK, limited to reduced fentanyl clearance after the final exposure. Pronounced sex differences in fentanyl metabolism were observed with female animals clearing both fentanyl and norfentanyl more slowly and producing less norfentanyl than males.

Introduction

Fentanyl is a potent synthetic opioid that was first approved by the United States Food and Drug Administration (FDA) in 1968 for induction of general anesthesia and the treatment of perioperative pain.¹ Since then, its approval has been extended to treat complex pain conditions, as its analgesic potency is about 50–75 times higher than morphine depending on the context and route of administration.^2–5 Unfortunately, fentanyl has now become a commonly used recreational drug.⁶ This increase in illicit use coincides with a rising number of overdose deaths involving fentanyl.^{7, 8} After intravenous fentanyl injection the pharmacokinetic (PK) profile is described by rapid redistribution followed by slow terminal elimination.⁹ In humans, the terminal half-life (t_1/2) ranges between 4–15 hours depending on age, whereas in rats the t_1/2 is significantly shorter (44 to 75 minutes) depending on the PK model.^10–12 Due to its lipophilicity and large volume of distribution, fentanyl has a long context-sensitive half-life, increasing the risk of accumulation and overdose if injected repeatedly within a short period of time.¹³

Fentanyl is primarily metabolized in the liver via cytochrome P450 3A4 (CYP3A4) to the inactive compound norfentanyl.^{14, 15} Less than 1% is metabolized to hydroxyfentanyl, hydroxynorfentanyl, and despropionylfentanyl and about 10% of the intact molecule is excreted renally and via feces.¹⁶ Fentanyl itself does not seem to alter CYP3A4 activity but CYP3A4 inducers or inhibitors (e.g. ketoconazole) influence fentanyl plasma concentrations.^{17, 18} A recent study using a physiologically-based PK (PBPK) model concluded that only 33% is metabolized to norfentanyl, while previous work suggested that it was greater than 80%.^{15, 16, 19, 20} There is limited knowledge regarding the influence of repeated fentanyl exposures on its metabolism, and the role of PK in the development of tolerance. However, in chronic opioid users the renal clearance is significantly protracted.^{21, 22} Since there are sex-specific differences in cytochrome P450 expression, fentanyl metabolism may differ between males and females.²³ However, most studies evaluating sex differences in opioid pharmacology have focused on pharmacodynamic (PD) and not PK parameters.^{24, 25} Even though there are previous studies investigating sex differences in morphine metabolism and antinociception in rats, these findings cannot be extrapolated to fentanyl since morphine only undergoes phase II metabolism.^{26, 27}

Therefore, in this study we evaluated sex differences in fentanyl metabolism and the effect of repeated high-dose fentanyl exposures on PK. We measured both fentanyl and norfentanyl plasma concentrations in an established rat model of repeated overdose to mimic illicit fentanyl use.²⁸ Collected concentration data were then analyzed using multiple PK approaches to assess the influence of sex and repeated exposures on fentanyl PK.

Methods

The Massachusetts General Hospital Institutional Animal Care and Use Committee approved all experimental procedures described in this study. Animals were treated following the Guide for the Care and Use of Laboratory Animals and all efforts were made to reduce the number of animals and to minimize suffering.²⁹ The report of the study follows the ARRIVE guidelines.³⁰

Animals

Animals were purchased from Charles River Laboratories (Wilmington, MA, United States) with an indwelling arterial catheter in the femoral artery connected to a vascular access button. Catheters were maintained by flushing weekly with heparinized glycerol (#HGS-10, SAI Infusion Technologies, Lake Villa, IL, USA). Adult male and female Sprague-Dawley rats (280–380 g) were single housed to avoid damage to the vascular access buttons by cage mates and had access to food and water ad libitum. They were kept on a 12-hour light on/light off cycle (7 am/7 pm) and a temperature of 22 °C. After arriving at the animal facility, the animals had at least one week to acclimatize before the start of the experiment.

Experimental design

The main purpose of this study was to investigate the influence of repeated high doses of fentanyl on PK. Additionally, we aimed to evaluate the influence of sex on these parameters. Since the literature on these topics is limited, we could not perform a proper sample size calculation and therefore relied on group sizes previously used to assess sex differences in opioid tolerance and PK in rats.^{26, 31} The group sizes ranged from 3–8 animals, so we included 6–7 animals per sex in our study.

We identified 100 μg/kg body weight as an intravenous dose in rats that reliably induced prolonged unconsciousness (assessed by the loss of righting reflex) mimicking human overdose. This dose was previously used in non-ventilated rats to evaluate the effect of fentanyl on respiratory drive and no fatalities were reported.³² Since fentanyl is commonly administered intravenously, we established a tail vein catheter before each fentanyl injection.³³ To place the catheter, we put the animals in an acrylic chamber and induced general anesthesia with isoflurane (5 Vol.-%). After loss of righting reflex, animals were transferred to a heating pad and anesthesia was maintained through continuous isoflurane inhalation via face mask (2–2.5 Vol.-%). A 24-gauge intravenous catheter was inserted in the tail vein, flushed with saline, and secured with tape. Animals were transferred to the home cage and had at least 45 min to recover from the procedure before fentanyl was administered.

Based on the results of a previous study investigating the effect of fentanyl tolerance on respiratory depression, we injected the animals twice a week with at least 72 hours in between exposures to ensure full washout of fentanyl and to provide sufficient time to recover.³⁴ All fentanyl injections were administered between 10 am and 12 pm.

Sample collection

Samples were collected from the arterial catheter after the first, fifth, and tenth injections as we observed that the time to return of righting reflex was longest after the first exposure, shortest after the fifth exposure, and then gradually increased again.²⁸ For four animals, samples were collected after the 11th injection, as blood had already been drawn from these animals at the 10th injection for blood gas analysis, in line with the 3Rs principles to minimize animal use. Blood was drawn 3, 15, 30, 60, 180, and 360 min after the injection based on previous studies investigating the PK of fentanyl.¹¹ Animals that died before the collection of the 360 min-sample were excluded from analysis. Accounting for the arterial catheter’s dead volume of approximately 0.09 ml, 0.15 ml of blood was first drawn and discarded. For analysis, a total of 0.30 ml of blood was drawn using a 1 ml-syringe pre-rinsed with ethylenediaminetetraacetic acid (EDTA). Afterwards, the arterial catheter was flushed with 0.25 ml of 5 IU/ml heparin/saline solution (Shenzhen Techdow Pharmaceutical Co., Shenzhen, China). To prevent clotting and to maintain patency of the arterial catheter, it was locked with 0.1 ml of heparin/glycerol solution (500 IU/ml) after the 60, 180, and 360 min blood draws. The blood was transferred into an EDTA tube (#B-D365974Z, Medline Industries Inc., Northfield, IL, USA) which was immediately placed on ice. The sample was centrifuged for 15 min at 1200 x g at 4°C. After centrifugation, plasma was isolated and stored at −80°C.

Sample analysis

Quantifiying fentanyl and norfentanyl concentrations in rat plasma (EDTA as anticoagulant) utilized protein precipitation followed by ultra-high-performance liquid chromatography (UHPLC) - electrospray ionization (ESI) - tandem mass spectrometry (UHPLC-ESI-MS/MS). Fentanyl-d5 and Norfentanyl-d5 were used as the internal standards. All reference materials were obtained from Cerilliant (Round Rock, Texas, USA). A 50 μL volume of the calibration standards, QCs, blanks, and samples was aliquoted, followed by addition of 50 μL of internal standard solution (1 ng/mL). Cold acetonitrile (500 μL) was added to precipitate plasma proteins, followed by centrifugation to separate the proteins, and then 500 μL of the supernatant was transferred to a fresh sample tube. The supernatant was evaporated to dryness, and the extract reconstituted in 100 μL of 10:90 methanol:0.1% aqueous formic acid. After mixing, the reconstituted extracts were transferred to autosampler vials for analysis by UHPLC-ESI-MS/MS. The UHPLC system was a ThermoScientific (San Jose, CA) Vanquish UHPLC autosampler and pumping system interfaced with a ThermoScientific TSQ Quantis Plus triple quadrupole mass spectrometer. Chromatographic separation was performed on a Phenomenex (Torrance, CA) Kinetex^® Biphenyl column (2.1 × 50 mm, 1.7 μm particle size) using gradient elution flowing at 0.3 mL/min. The mobile phases were (A) 0.1% aqueous formic acid and (B) methanol, initially equilibrated at 10% B. The mobile phase composition was linearly increased to 100% B at 2.5 min, where it was held for 1 min before being returned to and equilibrated at initial conditions. The MS/MS was operated in the Selected Reaction Monitoring (SRM) mode under positive ESI conditions. For fentanyl, fentanyl-d5, norfentanyl, and norfentanyl-d5, the SRM transitions were mass/charge (collision energy) 337.2 to 188.2 (23 eV), 342.2 to 188.2 (23 eV), 233.2 to 84.2 (19 eV), and m/z 238.2 to 84.2 (19 eV), respectively. Calibrators were prepared for each analytical batch in rat plasma at nominal fentanyl and norfentanyl concentrations between 0.01 and 10 ng/mL. Samples determined to have concentrations >10 ng/mL were re-assayed with either a 10- or 20-fold dilution into the range of the assay. A double blank (matrix without internal standard) and zero sample (matrix with internal standard) were prepared for each run in the same lot of rat plasma used for the calibrators. Quality Control samples (QCs) were also prepared in rat plasma at nominal fentanyl/norfentanyl concentrations of 0.03, 0.2, 2, and 8 ng/mL. Inter-day accuracy (as % difference from nominal concentration) across all QC concentrations was within ±10.3% for fentanyl and ±4.6% for norfentanyl, while inter-day precision (as % coefficient of variation, %CV) was within 4.9% for fentanyl and 5.2% for norfentanyl.

PK analysis

Multiple approaches for PK analysis were utilized to assess the role of sex and repeated exposures on the PK of fentanyl. First, fentanyl and norfentanyl concentration data were analyzed using a non-compartmental approach using Certara Phoenix WinNonlin^® v8.3.2. Elimination slopes were best-fit by the software, and the linear trapezoidal-linear interpolation method was used for calculations. Statistical differences between the calculated PK parameters (area under the concentration time curve to infinity (AUC_∞), maximum concentration (C_max), and half-life (t_1/2)) were assessed using two-way ANOVA with a Tukey multiple comparison test in GraphPad Prism v11.0.1. Next, we constructed a population PK model using NONMEM (v7.5.1) interfaced with Pirana (v2.9.9) to determine covariates significantly associated with the PK of fentanyl. The first-order conditional estimation with interaction was utilized throughout model construction. Model selection was based on parsimony, objective function value (OFV, representing the −2 log likelihood), and diagnostic plots. The ADVAN13/TRANS1 built-in model/parameterization subroutines were utilized to enable construction of a parent-metabolite model using user-written differential equations. After the base model was established, covariates (including age (weeks), weight (grams), sex (male vs. female), and injection number (1, 5, or 10) were tested on the model using the stepwise-covariate modeling (SCM) script available in Pirana. The SCM script tests covariate significance using forward inclusion (p<0.05), followed by backwards exclusion (p<0.01), using a χ² test with 1 degree of freedom. The final full model was then evaluated using a prediction corrected visual predictive check and bootstrapping, both using 1000 simulated datasets and scripts available in Pirana.

Finally, we developed a rat PBPK model in PK-Sim v11.2 to investigate the physiologic rationale behind the observed alterations in PK. A 17-compartment PBPK model representing virtual “organs” interconnected by blood flow was used for all simulations. The model was adapted from two human PBPK models; Kovar et al and an unpublished model developed to guide dosing in critically ill children.¹⁹ Rat specific model parameters were obtained from two rat models published by Zhang et al. and Shao et al.^{45, 46} Tissue specific organ partitioning coefficients were obtained from the rat models. Human models were used to determine the fraction of fentanyl metabolism for each specific CYP (CYP3A4/5 in humans). Including CYP3A4/5 specific metabolism allowed for the scaling of the model to rat and inclusion of the metabolite norfentanyl.

The rat fentanyl model was built using orthologs for human CYP3A4/5 (CYP3A18, CYP3A1/23, CYP3A62, and CYP3A9) and rat specific physiology. Rat profiles within the modeling software did not include sex specific physiology. Sex specific physiological parameters were taken from the literature to create separate male and female rat individuals (Table 2).^35–41 Size differences were accounted for within the software through calculated adjustments to organs based on inputted weight values. Additional changes were then made to reflect sex specific differences that could not be accounted for based on size alone. Expression data for rat CYPs were pulled from Bgee.org version 15.2.0 (Université de Lausanne). Sprague-Dawley expression data were selected when possible. If not present, the mean value of post pubescent or adult values were used. CYP data were included for liver, kidney, and gastrointestinal tract. Sex specific expression profiles were created to represent the difference in expression between male and female rats. Male and female individuals were created for model optimization using mean values for weights. Fentanyl metabolism was then divided among CYPs based on rat microsomal in vitro studies.^{15, 47, 48} Sex specific individual CYP V_max and K_m constants were estimated based on in vitro data and optimized using a Monte Carlo based algorithm within PK-Sim. K_cat values were then calculated within the software based on individual CYP abundance and V_max. CYP3A62 was included in initial models, but removed from subsequent models due to negligible contributions to fentanyl metabolism and model performance.⁴⁷ Final parameter optimization was completed using a Monte Carlo based algorithm within PK-Sim^® (Table 3).

Table 2:

PBPK model parameters used for male and female individual rats. Sex-specific physiologic parameters were adjusted based on published data.^{19, 35–38, 45, 46}

Parameter	Male	Female
Weight (g)	374	328
Venous Blood (mL)	17.54	17.8
Arterial Blood (mL)	7.63	7.8
Muscle (L)	0.02	0.01
Fat (L)	0.06	0.08
Hematocrit	0.44	0.4

Table 3:

Fentanyl and Norfentanyl PBPK model parameters

PARAMETER	FENTANYL	NORFENTANYL
Physicochemical properties
Lipophilicity (LogP)	3.95 *	2.68*
Plasma binding protein	Albumin19	Albumin19
Fraction unbound (Fu) (%)	10*	97*
Molecular weight (g/mol)	336.4719	232.3219
pKa	8.9919, 45	10.0019, 45
Distribution
Partition coefficients	Rodgers and Rowland19	Schmitt19
Cellular permeabilities	Charge-dependent Schmitt19	PK-Sim Standard19
Metabolism and Elimination
GFR	1.019
Plasma clearance renal (mL/min/kg)		16.4*
CYP450 Enzymes
	Male	Female
CYP3A1/23
Km (uM)	3715, 47, 48	3715, 47, 48
Kcat (L/min)	8.00c	1.00c
CYP3A18
Km (uM)	3715, 47, 48	3715, 47, 48
Kcat (L/min)	1.86c	57.14c
CYP3A9
Km (uM)	3715, 47, 48	3715, 47, 48
Kcat (L/min)	9.38c	0.16c

^*Values were optimized within modeling software

^cK_cat was calculated in the modeling software using V_max and CYP abundance

After optimization of fentanyl parameters, a norfentanyl compound was created, and integrated into the model as a metabolite. Parameters for norfentanyl were obtained from the Kovar et al. and Zhang et al. models.^{19, 46} Male and female individuals were then created using mean values for weights across all injection days to capture changes in growth during the experimental timeframe. These individuals were used as the mean for virtual populations. Two populations were created; 1 female and 1 male. Each population contained 1,000 virtual rats with variation in CYP enzyme expression of +/− 62% was added to account for biological variation. Dosing time was included as the median time within the experimental window of approximately 4 hours. Model performance was then evaluated using a published rat fentanyl and norfentanyl PK study.⁴⁹ A sensitivity analysis was then run in PK-Sim to determine parameter importance of sex specific differences to drug exposure (AUC). Variabilities were set to +/−0.1, 0.5, 1, 2.5, 10 fold variability.⁴⁵ All sex-specific parameters in Table 2 were tested. Analyses and visualizations were completed with R Version 4.6.1 using tidyverse.⁵⁰

Results

Study Population

A total of 19 animals were included in the present study (9 females and 10 males), to ensure data for 7 male and 7 female rats were available for each time point. Animals were replaced in the study due to death following the first injection (n=2), loss of arterial catheter patency during the study (n=2), or other health issues during the study period (n=1). Female rats were older (13.4 vs. 8.9 weeks; p<0.0001) and smaller (286 vs. 343 g; p<0.0001) than males at baseline.

Non-Compartmental Analysis of Fentanyl and Norfentanyl

Figure 1 depicts the concentration-time profiles of fentanyl and norfentanyl. Concentration–time profiles of fentanyl and norfentanyl were similar across injections, with no statistically significant changes across repeated doses. The t_1/2 of fentanyl observed in this study (Supplemental Table 1) indicates that substantial drug accumulation is unlikely with the 72-hour dosing interval used in the multi-dose protocol. Additionally, no statistically significant differences in fentanyl area under the concentration-time curve to infinity (AUC_∞) or maximum concentrations (C_max) were observed between male and female animals (figure 2). In contrast to this finding, norfentanyl AUC_∞ and C_max were lower (p<0.0001) in females, relative to males (figure 2). Female sex was also associated with a longer norfentanyl t_1/2 than males (Figure 2).

Figure 1:

Plasma concentration-time profiles of both (a) fentanyl and (b) norfentanyl. Data points represent mean (standard deviation) concentrations from 14 (all, in gray) or 7 (female, in red; males, in blue) rats. Data from the first injection are denoted with a filled circle, the fifth injection is marked with a filled square, and the tenth injection is depicted with a filled triangle.

Figure 2:

(a-c) Fentanyl and (d-f) norfentanyl AUC_∞, C_max, and t_1/2, respectively, separated by sex and injection number. Data represent mean (standard deviation) concentrations from 7 rats. Two-way ANOVA with Tukey multiple comparison tests were used for statistical analyses: * 0.05≥p>0.01; ** 0.01≥p>0.001; *** 0.001≥p>0.0001; **** p≤0.0001.

Population PK Analysis of Fentanyl and Norfentanyl

We then constructed a population PK model to identify the PK process underlying the sex-associated differences in norfentanyl exposure, as well as to assess the sex-based weight and age differences in the study animals. The final population PK model schematic and code are included in the supplements (Supplemental Figure 1 and Supplemental Material 1). Fentanyl PK was best described using a two-compartment model, while norfentanyl PK was best described with a one-compartment model, both parameterized using clearance and volume. Norfentanyl production was driven via Michaelis-Menten kinetics, as this approach yielded an extremely significant reduction in OFV relative to the use of a linear clearance model (ΔOFV=−161, p<0.0001). Inter-individual variability was modeled using η-values on fentanyl clearance (CL_Fent), the maximum velocity of norfentanyl production (V_max), and norfentanyl clearance (CL_Nor); attempts to estimate η-values on other parameters (including volumes) resulted in poor performance and instability in the model. Both fentanyl and norfentanyl utilized proportional error models.

Covariate analysis identified sex (p=0.001) and injection number (p=0.004) on CL_Fent, sex on V_max (p<0.0001), and sex on CL_Nor (p=0.004) as model significant. Of note, while injection number was categorical during covariate modeling, the parameter estimate describing the difference between injection one and five was minimal, so this covariate was dichotomized (injection 10 vs. injection 1 and 5) prior to further analysis with the model, with minimal change in the OFV (ΔOFV=0.774). These results indicated that CL_Fent is 16% slower in females relative to males and 8% slower after the 10^th injection as compared to injection 1 or 5. Production of norfentanyl (V_max) is reduced by 76% and CL_Nor is reduced by 45% in females relative to males. Final parameter-covariate equations and parameter estimates are included in Table 1, in addition to results of the bootstrap analysis. The goodness of fit plots and visual predictive check (Supplemental Figure 2) support model acceptance, as do the bootstrap results (Table 1), though there is some indication in these results of model instability.

Table 1:

Final population PK model parameter estimates and bootstrap analysis results.

Parameter	Estimate (%RSE)[Eta Shrinkage]	Bootstrap (95% CI)
CLFent (mL/hr)	774 (5%)	771 (689, 858)
CLFent:Sex	−0.158 (26%)	−0.157 (−0.238, −0.0787)
CLFent:Injection	−0.0796 (54%)	−0.077 (−0.165, 0.00594)
VFent,Central (mL)	275 (9%)	274 (219, 332)
QFent (mL/hr)	232 (18%)	254 (102, 363)
VFent,Peripheral (mL)	326 (9%)	335 (251, 401)
Km (ng/mL)	2.89 (17%)	3.19 (0.557, 5.22)
Vmax (ng/mL/hr)	78.8 (12%)	71.5 (10.0, 148)
Vmax:Sex	−0.764 (7%)	−0.761 (−0.881, −0.647)
CLNor (mL/hr)	23.6 (11%)	21.0 (4.25, 42.9)
CLNor:Sex	−0.445 (27%)	−0.434 (−0.728, −0.162)
VNor (mL)	10.4 (18%)	9.23 (2.84, 18.0)
ωCL,Fent2	0.00843 (20%) [9%]	0.00648 (0.00300, 0.0138)
ωVmax2	0.0716 (37%) [11%]	0.0536 (0.00794, 0.135)
ωCL,Nor2	0.0602 (65%) [24%]	0.0387 (−0.00988, 0.130)
Prop. Error, Fentanyl	0.112 (17%) [2%]	0.107 (0.0792, 0.145)
Prop. Error, Norfentanyl	0.261 (11%) [3%]	0.261 (0.206, 0.316)

Abbreviations: CL_Fent, Fentanyl Clearance; CL_Fent:Sex, Sex covariate effect on Fentanyl Clearance; CL_Fent:Injection, Injection covariate effect on Fentanyl Clearance; V_Fent,Central, Fentanyl Central Compartment Volume; Q_Fent, Fentanyl Intercompartmental Clearance; V_{Fent,Peripheral}, Fentanyl Peripheral Compartment Volume; K_m, Michaelis Constant; V_max, Maximum Enzyme Reaction Rate; V_max:Sex, Sex covariate effect on Maximum Enzyme Reaction Rate; CL_Nor, Norfentanyl Clearance; CL_Nor:Sex, Sex covariate effect on Norentanyl Clearance; V_Nor, Norfentanyl Volume; ω_CL,Fent², Inter-Individual Variability on Fentanyl Clearance; ω_Vmax², Inter-Individual Variability on Maximum Enzyme Reaction Rate; ω_CL,Nor², Inter-Individual Variability on Norfentanyl Clearance.

PBPK Analysis of Fentanyl and Norfentanyl

A rat fentanyl and norfentanyl PBPK model was developed to identify a physiological rationale for the observed sex-specific differences in fentanyl/norfentanyl Sex-specific differences were incorporated into the virtual populations and model parameters (Table 2 and Table 3). Male and female model simulations were run separately for fentanyl and norfentanyl by dosing regimen (Figure 3, Supplemental Figure 3). Additional model evaluation was performed by comparing model estimates with a published PK study in male rats (Supplemental Figure 4).⁴⁹ Model performance varied by sex and study day. Fentanyl model performance was best on day 1 with 93% of male and 97% of female observed data within the prediction interval. By day 31‒36, about 80% of male observed data fell within the prediction interval, however, model performance remained high in females with 93% of observed data within the prediction interval. Female norfentanyl model predictions performed best and remained above 90% for the entirety of the experimental timeframe. Male norfentanyl model performance ranged from 82% of observed data within the prediction interval on day 1 to 60% of observed data falling within the prediction interval by days 31‒36. External model evaluation using published pharmacokinetic data from male Wistar rats demonstrated good agreement for norfentanyl (Supplemental Figure 4).

Figure 3:

Fentanyl and norfentanyl PBPK model evaluations for male and female rats on Day 1. Black circles represent individual plasma concentrations, the black line is the median model concentration prediction, and the grey shaded area represents the 90% confidence interval.

A sensitivity analysis was performed for male and female PBPK model simulations to identify key sex-specific physiologic parameters important in fentanyl PK (Supplemental Figure 5). A 10-fold increase in sex-specific parameters values was used to assess the relative impact on the PK of fentanyl and norfentanyl. Muscle volume, fat volume, hematocrit and venous blood volume were all identified as important, sex-specific parameters for total clearance, C_max and AUC_inf (Supplemental Figure 5). In both sexes, fentanyl PK was sensitive to alterations in muscle volume, venous blood volume, hematocrit, and fat volume. However, muscle volume and hematocrit had a greater effect in males with fat volume more pronounced in females. For norfentanyl, hematocrit had the largest effect on PK in both males and females. Overall, more sex specific physiological parameters were associated with the PK of fentanyl than norfentanyl.

Discussion

In this study, we evaluated the effect of repeated, high-dose fentanyl exposures on its metabolism using a rodent model. The non-compartmental analysis revealed no significant differences in fentanyl metabolism over ten injections. However, there were sex-specific differences in its metabolism as norfentanyl AUC_∞ and C_max were lower, while norfentanyl t_1/2 was longer in the females compared to males. Constructing a population PK model, we found that fentanyl followed best a two-compartment model whereas norfentanyl was best described by a one-compartment model. The covariate analysis confirmed the significant influence of sex on fentanyl metabolism, revealing that fentanyl clearance, norfentanyl clearance, and norfentanyl production were significantly slower in females. Additionally, overall fentanyl clearance was significantly slower after the 10^th exposure. The PBPK model performed well for fentanyl but less accurately for norfentanyl. The sensitivity analysis identified hematocrit and plasma protein concentrations as key physiological factors affecting fentanyl and norfentanyl kinetics in both sexes. For fentanyl, fat volume was positively correlated with drug exposure in females, while in males greater muscle volume increased fentanyl clearance.

We found sex-specific differences in fentanyl metabolism which might be explained by species-specific metabolism and sex-specific expression of cytochrome P450 enzymes. In humans, fentanyl is primarily metabolized to norfentanyl by CYP3A4.¹⁵ However, in rats the major CYP isoform catalyzing fentanyl oxidation is not its human ortholog CYP3A9, but rather CYP3A1/2.^{47, 51} Whereas CYP3A9 is induced by estrogen and has significantly higher gene expression in female hepatic samples than in males, CYP3A2 is a male-specific isoform of cytochrome P450.^{23, 52} For these reasons, we observed an increased fentanyl clearance and faster norfentanyl production in males compared to females. In contrast to our findings, Ohtsuka et al. did not identify statistically significant sex differences in fentanyl pharmacokinetics following intravenous administration in rats but reported a non-significant trend toward higher fentanyl clearance in females.³¹ The reasons for this discrepancy may relate to differences in study design, including the lower fentanyl dose, the small sample size, and the use of a one-compartment pharmacokinetic model. Furthermore, the present study included norfentanyl measurements and utilized population and physiologically based pharmacokinetic modeling approaches, which may have increased sensitivity to detect sex-dependent differences in fentanyl metabolism and clearance. Interestingly, overall fentanyl exposure did not differ significantly between sexes, suggesting that females may eliminate a greater proportion of fentanyl through non-hepatic pathways. Previous studies have focused on sex-specific differences in fentanyl pharmacodynamics rather than PK.^{24, 25} Although findings vary depending on the endpoint assessed, most studies reported no significant sex differences in fentanyl-induced analgesia, respiratory depression, or sedation.^{28, 34, 53, 54} However, most studies primarily evaluated peak pharmacodynamic effects, whereas the principal sex-dependent PK difference observed in the present study was fentanyl clearance, which may have a greater influence on the duration rather than the magnitude of fentanyl effects. Nevertheless, Bartock et al. did not identify significant sex differences in either the magnitude or time course of fentanyl-induced antinociception in rats.⁵³ The observed sex differences in norfentanyl formation are also unlikely to translate into pharmacodynamic effects, as norfentanyl is considered pharmacologically inactive. Taken together, the available pharmacodynamic literature in rodents is heterogeneous and does not support a consistent pattern of sex-dependent differences in fentanyl effects. This is broadly consistent with our findings, which demonstrate sex differences in fentanyl metabolism but not in overall fentanyl exposure.

The changes in fentanyl metabolism with repeated exposures were limited, but we observed a decreased fentanyl clearance after the 10^th exposure. This is in line with a previous study that demonstrated a protracted renal clearance of fentanyl in persons with opioid use disorder (OUD).²¹ A potential mechanism might be the downregulation of CYP enzyme expression. Opioids themselves can influence CYP expression, as it was shown that morphine downregulates CYP3A1, which is known to be involved in fentanyl metabolism in rats.⁵⁵ Additionally, hypoxia-inducible factor (HIF) activity has been linked to suppression of CYP3 family expression.⁵⁶ In our model the rats were repeatedly hypoxic, which causes HIF expression, and fentanyl itself activates HIF in a μ-opioid receptor (MOR)–dependent manner. Therefore, HIF-associated downregulation might explain the reduced fentanyl clearance with repeated exposures.⁵⁷ However, the fentanyl tolerance observed clinically in individuals with OUD indicates that these effects may be less relevant in humans, or are outweighed by adaptations in MOR signaling.⁵⁸ In opioid-naïve individuals, Egan et al. reported no meaningful changes in fentanyl pharmacokinetics following repeated, short-interval transmucosal fentanyl administration.⁵⁹ Despite important differences in study design, including species, route of administration, dosing regimen, and the absence of complete washout between doses, these findings are consistent with our results. Although both male and female participants were included in that study, sex-specific pharmacokinetic analyses were not reported, limiting direct comparison with the sex-dependent findings observed in the present study.

Consistent with previous reports, fentanyl PK was best captured by a two-compartment model, reflecting its high lipophilicity, rapid distribution into well-perfused tissues, and prolonged redistribution from peripheral compartments.⁶⁰ As a consequence, the t_1/2 observed in our study aligns with values reported in two-compartment analyses but exceeds those derived from one-compartment models, which underestimate the slower terminal elimination phase.^{12, 31, 61} Our PBPK model underlines the role of redistribution from peripheral compartments such as muscle and fat as key determinants of fentanyl clearance. We found that muscle volume determined total body clearance in males, whereas fat volume increased overall drug exposure in females, which is presumably attributable to sex differences in body composition.⁶² Since redistribution from muscle occurs more rapidly than from fat, not only the previously described sex differences in CYP enzyme expression but also the higher percentage of body fat in females may contribute to the longer fentanyl clearance observed in this group.⁶³

Limitations

In the present study we only evaluated PK following IV fentanyl administration. We focused on this route of administration since it is one of the predominant routes in illicit fentanyl use and is most commonly associated with overdose deaths.^{33, 64} As we were interested in the effects of repeated fentanyl overdose we chose a bolus dose of 100μg/kg which was associated with a lethality of approximately 10% in opioid naïve animals.²⁸ Consequently, two animals died following the first fentanyl injection, and a survivorship bias cannot be excluded, as their pharmacokinetic profiles may have differed from those of the surviving animals included in the analysis. To ensure a full washout of fentanyl in between exposures and give the animals enough time to recover between doses we administered fentanyl twice per week with at least 72 hours in between exposures.

Additionally, our findings are derived from a rat model designed to mimic repeated fentanyl overdoses. Given the limitations of animal models in predicting human toxicity and known species-specific differences in fentanyl metabolism, the direct translatability of these results to humans may be limited.⁶⁵ Furthermore, only norfentanyl was quantified in plasma in the present study. Although norfentanyl represents the major metabolic pathway of fentanyl, neither other metabolites nor urinary excretion were assessed.^{14, 15} Therefore, the contribution of alternative metabolic pathways or renal clearance to the observed sex differences could not be determined.

Conclusion

In this study, we evaluated the impact of repeated high-dose fentanyl administration on its PK. In addition to a modest reduction in fentanyl clearance after the final exposure, we identified pronounced sex differences in fentanyl metabolism. Female animals exhibited slower clearance of both fentanyl and norfentanyl, and they produced overall less norfentanyl than males. These findings highlight the importance of sex-specific analysis and warrant future studies to evaluate potential clinical implications.

Significance statement.

In a rat model of repeated fentanyl exposure mimicking illicit use, pharmacokinetic changes over time were minimal, suggesting limited accumulation with intermittent dosing. In contrast, marked sex differences were observed, with female rats showing slower fentanyl clearance and reduced metabolite formation, highlighting sex-specific determinants of opioid metabolism with potential implications for interpreting toxicology findings and overdose risk.

Abbreviations

AUC_∞

Area under the concentration–time curve extrapolated to infinity

CL

Clearance

C_max

Maximum plasma concentration

EDTA

Ethylenediaminetetraacetic acid

ESI

Electrospray ionization

HIF

Hypoxia-inducible factor

K_m

Michaelis constant

MOR

μ-opioid receptor

OFV

Objective function value

OUD

Opioid use disorder

PBPK

Physiologically based pharmacokinetic

PD

Pharmacodynamic

PK

Pharmacokinetic

QC

Quality control

SCM

Stepwise covariate modeling

SRM

Selected reaction monitoring

t_1/2

Elimination half-life

UHPLC

Ultra-high-performance liquid chromatography

UHPLC-ESI-MS/MS

Ultra-high-performance liquid chromatography–electrospray ionization–tandem mass spectrometry

V_max

Maximum enzyme reaction rate

Data Availability Statement

The authors declare that all the data supporting the findings of this study are contained within the paper.