Physiologically Based Pharmacokinetic Modeling: The Reversible Metabolism and Tissue-Specific Partitioning of Methylprednisolone and Methylprednisone in Rats

Abstract

The pharmacokinetics (PK) of methylprednisolone (MPL) exhibited tissue-specific saturable binding and reversible conversion with its metabolite, methylprednisone (MPN). Blood and 11 tissues were collected in male rats after intravenous (i.v.) bolus doses of 50 mg/kg MPL and 20 mg/kg MPN and upon i.v. infusion of MPL and MPN at 0.3, 3, and 10 mg/h per kg. The concentrations of MPL and MPN were simultaneously measured. A comprehensive physiologically based pharmacokinetic (PBPK) model was applied to describe the plasma and tissue profiles and estimate PK parameters of the MPL/MPN interconversion system. Both dosed and formed MPL and MPN were in rapid equilibrium or achieved steady-state rapidly in plasma and tissues. MPL tissue partitioning was nonlinear, with highest capacity in liver (322.9 ng/ml) followed by kidney, heart, intestine, skin, spleen, bone, brain, muscle, and lowest in adipose (2.74 ng/ml) and displaying high penetration in lung. The tissue partition coefficient of MPN was linear but widely variable (0.15∼5.38) across most tissues, with nonlinear binding in liver and kidney. The conversion of MPL to MPN occurred in kidney, lung, and intestine with total clearance of 429 ml/h, and the back conversion occurred in liver and kidney at 1342 ml/h. The irreversible elimination clearance of MPL was 789 ml/h from liver and that of MPN was 2758 ml/h with liver accounting for 44%, lung 35%, and kidney 21%. The reversible metabolism elevated MPL exposure in rats by 13%. This highly complex PBPK model provided unique and comprehensive insights into the disposition of a major corticosteroid.

SIGNIFICANCE STATEMENT

Our dual physiologically based pharmacokinetic (PBPK) study and model of methylprednisolone/methylprednisone (MPL/MPN) with multiple complexities reasonably characterized and parameterized their disposition, and provided greater insights into the interpretation of their pharmacodynamics in rats. Drug knowledge gained in this study may be translatable to higher-order species to appreciate the clinical utility of MPL. The complex model itself is instructive for advanced PBPK analysis of drugs with reversible metabolism and/or nonlinear tissue partitioning features.

Introduction

6α-Methylprednisolone (MPL) is a moderately potent glucocorticoid (GC) used widely in veterinary and human medicine. Like other therapeutic GCs, MPL manifests its anti-inflammatory and immunosuppressive effects in various inflammatory (e.g., spinal cord injury and asthma) and autoimmune diseases (e.g., rheumatoid arthritis and organ transplants) through genomic, nongenomic, and mitochondrial glucocorticoid signaling pathways by binding mainly to the ubiquitously expressed glucocorticoid receptor (GR) (Meduri and Chrousos, 2020). Additionally, MPL has superiority among GCs in the treatment of acute respiratory distress syndrome (ARDS) such as severe COVID-19 (Meduri et al., 2020; Hong et al., 2023).

6α-Methylprednisone (MPN), a biologically inactive metabolite of MPL, exhibits reversible conversion with its parent drug. The interconversion process between MPL and MPN is catalyzed by an endogenous steroid enzyme system, 11β-hydroxysteroid dehydrogenases (11β-HSDs), which appears to be widely distributed in liver, kidney, lung, placenta, and inflammatory cells (Li et al., 1997; He et al., 2016). Such interconversion has been partly assessed in rats and rabbits using moment analysis and compartment models (Ebling and Jusko, 1986; Haughey and Jusko, 1992). To be specific, 11β-HSD1 is a predominant reductase, mediating regeneration of MPL from MPN, whereas 11β-HSD2 is an exclusive dehydrogenase, inactivating MPL to MPN. A recently raised target-mediated drug disposition (TMDD) phenomenon for 11β-HSD1 complicates the reversible metabolism further (An and Katz, 2023). This is evidenced by the high-affinity and low-capacity binding of 11β-HSD1 and its inhibitor and the according concentration-time profiles with typical TMDD features (Wu et al., 2021).

A complete physiologically based pharmacokinetic (PBPK) model for prednisolone and prednisone was enacted in rats (Li et al., 2020). Nonlinear GC binding appeared in almost all tissues and implied the capacity-limited GC-GR interaction. MPL was poorly distributed into the central nervous system of pig and mouse (Koszdin et al., 2000; Bernards, 2006), owing to P-glycoprotein–mediated efflux transport at blood-brain barrier (BBB). An in situ liver perfusion study conducted in rats showed nonlinear hepatic extraction of MPL (Kong and Jusko, 1991). MPL exhibits better penetration into its pivotal target, lung, over prednisolone, and an exponential concentration increase of MPL in bronchoalveolar space was observed as plasma concentration increased (Vichyanond et al., 1989).

Considering the clinical utility of MPL and its various actions in tissues, we sought to characterize the tissue-specific drug disposition of the parent drug, MPL, and the metabolite, MPN. In this study, both parent and metabolite were administered and blood and tissue samples collected as required for pharmacokinetic (PK) analysis of the interconverting pair, MPL/MPN. A wide range of dosages was given via both bolus and zero-order infusions to assess the capacity-limited and steady-state features. A sensitive and specific liquid chromatography–tandem mass spectrometry (LC-MS/MS) assay method was applied to simultaneously measure MPL and MPN concentrations in all samples. Samples at later times than studied previously were included to capture the terminal phases. The PBPK model was assessed with various test configurations in the optimization process. Ultimately, a comprehensive whole-body PBPK model was developed to quantitatively depict drug kinetics in blood and tissues and to allow better interpretation of how tissue concentrations drive the diverse local effects of this corticosteroid.

Materials and Methods

Reagents and Chemicals

6α-Methylprednisolone (purity ≥98%) and 6α-methylprednisone (purity ≥98%) were purchased from Sigma-Aldrich (St. Louis, MO) and Steraloids (Newport, RI). Prednisolone [shared internal standard (IS) for MPL and MPN, purity ≥99%], liquid chromatography/mass spectrometry–grade acetonitrile, methanol, and high-performance liquid chromatography–grade formic acid were purchased from Sigma-Aldrich. Ethanol 200 proof (Decon Laboratories, King of Prussia, PA), dimethyl sulfoxide (DMSO; Sigma-Aldrich), polyethylene glycol 400 (PEG 400; Sigma-Aldrich), Captisol (CYDEX Pharmaceuticals, Lawrence, KS), 0.9% normal saline (VEDCO, St. Joseph, MO), and Milli-Q water (Millipore Corporation, Bedford, MA) were used as solvents.

Animals

Healthy male Wistar rats aged 5 to 6 weeks were purchased from Envigo (Indianapolis, IN) and housed two per cage under controlled conditions with free access to rat chow and drinking water in the University Laboratory Animal Facility. All rats were acclimated for at least 1 week before the experiments. The study protocols adhered to the Guide for the Care and Use of Laboratory Animals and were approved by the University at Buffalo Institutional Animal Care and Use Committee.

Pharmacokinetic Experiments

There were 51 rats in the PK study, weighing an average of 287 g with a range from 265 to 325 g. Both MPL and MPN were given in original drug form. The vehicle for bolus dosing solutions was made up of PEG 400, ethanol, and normal saline, 5:2:3 (v/v/v) for MPL and 6:2:2 for MPN; whereas a uniform vehicle was applied for MPL and MPN infusion solutions, 2% volume of DMSO: 5% ethanol: 33% Captisol: 60% normal saline, in which Captisol was solubilized in Milli-Q water in advance at 60%, w/v. All of the above solutions were prepared freshly and filtered through 0.22-μm filters before use. The drug was given i.v. in a volume of 4 ml/kg in the bolus study and 4 ml/h per kg for infusion.

The first group of 27 rats was given an i.v. bolus dose of 50 mg/kg MPL within 30∼60 seconds through the femoral vein and sacrificed by exsanguination at 0.25, 0.5, 1, 2, 4, 6, 8, 12, and 24 hours (n = 3 at each time point). Whole blood was drained from the abdominal aorta into EDTA-containing syringes. Half of the whole blood was centrifuged (4°C at 2000 g for 15 minutes) immediately to obtain plasma. Then 11 tissues, including heart, liver, spleen, lung, kidney, muscle (gastrocnemius), skin, visceral adipose, brain, bone (tibia), and intestine, were collected, rinsed in ice-cold PBS, and gently blotted, weighed, and immediately frozen in liquid nitrogen. All tissues, whole blood, and plasma samples were stored at −80°C before further analysis.

Similarly, a second group of six rats was randomly assigned into two subgroups and dosed with 20 mg/kg MPN i.v. over 90∼120 seconds. Then 200-μl blood samples were collected through the saphenous vein using EDTA-coated microcapillary tubes and centrifuged for plasma. The sampling timepoints were 0.25, 1, 4, 8, and 21 hours for rats in subgroup A and 0.5, 2, 6, and 12 hours for subgroup B.

An additional batch of 18 rats was involved in the MPL and MPN infusion studies, in which the two drugs were infused i.v. at 0.3, 3, and 10 mg/h per kg in three rats each. Rats were implanted with a jugular vein catheter for constant-rate infusion and a carotid artery catheter for serial blood sampling (at 0.25, 0.75, 1.5, 3, 5, and 7 hours from infusion start) via the Instech infusion system (https://www.instechlabs.com/applications/IV-drug-infusion/rat-continuous-infusion-blood-sampling). They were euthanized for tissue collections at the end of 7∼8 hours of infusion, when steady-states of MPL or MPN concentrations were expected. Tissue collection procedures were the same as those described in the bolus study.

Sample Preparation and MPL/MPN Concentration Analysis

Plasma, whole blood, and tissue samples were handled using a previously reported method (Li et al., 2020; Song et al., 2020) with slight modifications. Tissue samples were ground into fine powder under liquid nitrogen and homogenized in a bullet blender (Next Advance Inc., Troy, NY) with 2- or 3-fold volumes of cold PBS, whereas whole-blood samples were directly homogenized using a PRO-200 BIO-GEN homogenizer (PRO Scientific, Oxford, CT). Plasma or prepared homogenate totaling 100 μl was spiked with 10 μl IS working solution (prednisolone, 50 ng/ml), and drug was extracted by adding 500 μl acetonitrile containing 0.1% formic acid for protein precipitation. After vortexing and centrifugation (4°C at 13,000 g for 10 minutes), the supernatants were transferred to glass tubes and dried under nitrogen flow. The residue was then reconstituted with 100 μl methanol/water (30:70, v/v), vortexed, and centrifuged to obtain clear supernatant for MPL/MPN concentration analysis.

An LC-MS/MS method was established and validated to simultaneously determine the concentrations of MPL and MPN in all samples. The mobile phase consisting of eluent A (water containing 0.1% formic acid) and eluent B [methanol/water (95:5, v/v) containing 0.1% formic acid] was pumped at a flow rate of 0.2 ml/min with a gradient elution. The mass spectrometer was operated in the negative ion mode for the detection of ion transitions at mass-to-charge ratio 419.2/343.0 for MPL, 417.2/341.0 for MPN, and 405.2/328.9 for the IS. The lower limit of quantification was 0.1 ng/ml for both MPL and MPN in plasma, blood, and tissue homogenates. The LC-MS/MS assay was conducted in a system consisting of SCIEX ExionLC modules (Framingham, MA), including a binary pump, a degasser, an autosampler and a column oven, a SCIEX QTRAP 6500+ mass spectrometer (Singapore), and Analyst 1.7.2 software (Applied Biosystems SCIEX).

PBPK Model Development

The proposed dual PBPK model for the reversible MPL/MPN system is provided in Fig. 1. It consists of two sets of physiologic compartments assigned separately for MPL and MPN; in each there were two blood compartments (artery and vein), 11 tissues (liver, kidney, lung, heart, spleen, intestine, muscle, adipose, bone, skin, brain), and one remainder compartment that represents all other unmeasured tissues. This model was based upon several drug-specific observations and assumptions as follows: 1) Only free MPL and MPN in plasma partition into blood cells and tissues with instant equilibrium. 2) There is saturable tissue binding of MPL in almost all tissues (Ballard et al., 1974; Meduri and Chrousos, 2020), whereas the saturable binding of MPN happens only in liver (Tannin et al., 1991) and kidney (Krozowski, 1999). Only unbound drug in tissues undergoes conversion and elimination. 3) For MPL, the brain has two subcompartments (brain capillary and extravascular tissue) with bidirectional passive diffusion and unidirectional P-glycoprotein–mediated active efflux (Bernards, 2006). 4) Upon MPL or MPN administration, the appearance of the counterpart MPN or MPL in blood and tissues is due to MPL→MPN conversion within kidney (Krozowski, 1999), intestine (Zhou et al., 1995), and lung or MPN→MPL conversion within liver (Tannin et al., 1991) and kidney (Krozowski, 1999) and then systemic circulatory transport. 5) The irreversible elimination of MPL involves only hepatic metabolism (Czock et al., 2005), whereas MPN is eliminated through hepatic (Matabosch et al., 2013), renal (Panusa et al., 2011), and pulmonary routes.

Fig. 1.

Whole-body PBPK model scheme for methylprednisolone (MPL, left) and methylprednisone (MPN, right) disposition in rats. Lines with arrows indicate blood flows, drug administration, transport, conversion, and elimination. Symbols are defined in Tables 1–3.

TABLE 1.

Physiologic parameters of tissues and organs in 287-g rats

Tissue	Volume (V, ml)	Blood Flow (Q, ml/h)
Arterial blood	5.62a	5487.9b
Venous blood	11.25a	5487.9b
Liver	9.70a	1408.8b
Liver artery	Not applicable	39.4b
Lung	0.94a	5487.9b
Kidney	1.77a	679.6b
Heart	0.85a	281.9b
Intestine	7.95a	1136.3b
Spleen	0.42a	333.1b
Skin	50.93a	372.1b
Muscle	124.48a	1724.1b
Adipose	33.62a	418.2b
Bone	21.10a	114.4b
Brain (capillary, extravascular)	1.40 (0.02, 1.38)a	121.8b
Remainder	16.91c	367.2c

^aFrom Li et al. (2020).

^bFrom Shah and Betts (2012).

^cCalculated value 1) assuming 1 g/ml tissue density, volume for remainder = body weight − volume summation for listed tissues; 2) blood flow for remainder = cardiac output − blood flow for listed tissues.

TABLE 3.

Parameters for elimination and conversion processes of methylprednisolone (MPL) and methylprednisone (MPN) obtained in PBPK model fitting The units for CL, V_max, and K_m are ml/h, ng/h, and ng/ml.

Parameter	Definition	Estimate (CV%)
CLMPLe,liver	Elimination clearance of MPL in liver	1796 (10.3)
CLMPNe,liver	Elimination clearance of MPN in liver	8057 (19.6)
CLMPNe,kidney	Elimination clearance of MPN in kidney	4078 (17.1)
CLMPNe,lung	Elimination clearance of MPN in lung	1188 (11.9)
CLLN,kidney	MPL→MPN conversion clearance in kidney	558.7 (27.4)
Vmax_LN,lung	Maximal rate of MPL→MPN conversion in lung	196,700 (19.6)
Vmax_LN,intestine	Maximal rate of MPL→MPN conversion in intestine	136,000 (32.9)
Km_LN	Shared Michaelis-Menten constant of MPL→MPN conversion in lung and intestine	1319 (22.1)
CLNL,liverv	MPN→MPL conversion clearance in liver	2051 (39.1)
CLNL,kidney	MPN→MPL conversion clearance in kidney	2000 (48.1)

The model differential equations are listed below, with all initial conditions set to 0. The kinetics of MPL and MPN concentrations in heart, spleen, skin, muscle, adipose, bone, and the remainder compartment were:

in which V, Q, C, K_pu, and K_p are tissue volume, blood flow, drug concentration, unbound partition coefficient for tissues involving saturable drug binding, and partition coefficient for tissues without saturable drug binding. The blood-to-plasma ratio (R_b) was obtained using the measured blood (C_blood) and plasma (C_plasma) concentrations of samples from overall intravenous bolus and intravenous infusion studies in rats (i.e., R_b = C_blood/C_plasma). The subscripts denote the corresponding variables and parameters of indicated MPL or MPN in blood or tissues.

In liver, MPL and MPN concentration changes were:

in which CL_MPLe,liver, CL_MPNe,liver, and CL_NL,liver are the elimination clearance of MPL and MPN and MPN→MPL conversion clearance in liver and Q_liver = Q_liver,artery + Q_spleen + Q_intestine.

In kidney, MPL and MPN concentration changes were:

in which CL_MPNe,kidney, CL_LN,kidney, and CL_NL,kidney are the elimination clearance of MPN, MPL→MPN conversion clearance, and MPN→MPL conversion clearance in kidney.

In lung and intestine, MPL and MPN concentration changes were:

in which CL_MPNe,lung is the elimination clearance of MPN in lung, V_{max_LN,lung} and V_{max_LN,intestine} are the maximal rate of MPL→MPN conversion in lung and intestine, K_{m_LN} is the shared Michaelis-Menten constant of these two processes.

MPL concentration changes in brain capillary and extravascular brain and that of MPN in the entire brain were:

in which CL_{eff_MPL,brain} and CL_{up_MPL,brain} are the passive diffusion clearance and efflux clearance of MPL in brain.

Physiologically, drug in the venous system converges from all non-lung tissues and then perfuses the lung, whereas drug in the arterial system originates from lung and perfuses all of the other tissues. Therefore, MPL and MPN concentration changes of venous and arterial blood were:

in which Q_lung denotes the lung blood flow and is equal to the total cardiac output and K_{0_MPL} and K_{0_MPN} are the zero-order intravenous infusion rates of MPL and MPN.

For tissues involving nonlinear binding, their total tissue concentration (C_tissue) and unbound partition coefficient (K_pu) were:

in which B_max, K_d, and C_u,tissue are tissue-specific binding capacity, equilibrium dissociation constant, and unbound tissue concentration. Combining eqs. 18 and 19 yields:

Model Fitting

Naïve-pooled plasma and tissue concentration-time data from the entire PK study were analyzed jointly using the established whole-body PBPK model. In total, 39 PK parameters of the MPL/MPN interconversion system were estimated, including 14 tissue-specific binding capacities (B_max), 10 tissue/plasma partition coefficients (K_p), two brain uptake or efflux clearances of MPL (CL_{up_MPL,brain}, CL_{eff_MPL,brain}), three equilibrium dissociation constants (K_{d_MPL,lung}, K_{d_MPL,other}, K_{d_MPN}), six MPL/MPN interconversion parameters (CL_LN,kidney, V_{max_LN,lung}, V_{max_LN,intestine}, K_{m_LN}, CL_NL,liver, CL_NL,kidney), and four elimination clearances (CL_{MPLe_liver}, CL_{MPNe_liver}, CL_{MPNe_kidney}, CL_{MPNe_lung}). Physiologic parameters, including tissue volumes and blood flow rates for different tissues and organs, were fixed to literature values (Shah and Betts, 2012; Li et al., 2020) (Table 1). The inputs of the MPL and MPN intravenous bolus doses were treated as brief infusions with transient duration (45 seconds for MPL, 105 seconds for MPN). The output equation for plasma was described as the concentration in arterial blood compartments divided by R_b (i.e., C_plasma = C_artery/R_b). The measurements of MPL and MPN in all tissues were corrected for residual blood (Li et al., 2020; Song et al., 2020) and then used as the authentic observations.

The whole-body PBPK model fittings were implemented using ADAPT version 5 (Biomedical Simulations Resource, University of Southern California, Los Angeles, CA) (D’Argenio et al., 2009). The maximum likelihood estimation was applied for model fitting. The ADAPT code for this model is provided in the Supplemental Material. The variance model was set as V_i = (σ₁ + σ₂ • Y_i)², where V_i represents the variance of the ith data point, Y_i is the ith model prediction, and σ₁ and σ₂ are variance model parameters.

Results

Blood Partitioning of MPL and MPN

Figure 2 shows the relationships of blood-to-plasma ratio or blood partitioning (R_b) versus plasma concentrations of MPL and MPN. The R_{b_MPL} and R_{b_MPN} were constant over the concentration range of 0.1∼10,000 ng/ml with values of 0.64 ± 0.09 (n = 40) and 0.63 ± 0.07 (n = 35) based on the intravenous bolus and infusion studies and were applied as fixed values in the subsequent PBPK model. The blood partitioning of MPL and MPN is comparable to that of dexamethasone (R_b = 0.664) (Song et al., 2020) and prednisolone (R_b = 0.703) (Li et al., 2020) in male rats.

Fig. 2.

Blood-to-plasma ratios (R_b) of methylprednisolone (MPL, red circles) and methylprednisone (MPN, blue triangles) versus plasma concentration after intravenous bolus and intravenous infusion studies in rats. The dashed lines indicate the mean values of all data.

PBPK Model Fitting

The MPL/MPN concentration-time observations in arterial plasma and 11 collected tissues after the 50-mg/kg MPL i.v. bolus (Fig. 3), in venous plasma after the 20-mg/kg MPN i.v. bolus (Fig. 4), in arterial plasma during 0.3, 3, and 10 mg/h per kg i.v. infusion of MPL, in 11 tissues at steady-states (Fig. 5), and in arterial plasma during 0.3, 3, and 10 mg/h per kg i.v. infusions of MPN and in tissues at steady-state (Fig. 6) were jointly fitted using the whole-body PBPK model of MPL/MPN. Generally, the proposed model adequately captured almost all profiles, as reflected by the good agreements between the model-predicted and observed MPL/MPN concentrations across the eight different dosing sets.

Fig. 3.

Methylprednisolone (MPL, red) and methylprednisone (MPN, blue) concentration-time profiles in plasma and tissues after 50-mg/kg i.v. bolus doses of MPL in rats. Curves depict PBPK model fittings.

Fig. 4.

Methylprednisolone (MPL, red) and methylprednisone (MPN, blue) concentration-time profiles in plasma after 20-mg/kg i.v. bolus doses of MPN. Curves depict PBPK model fittings.

Fig. 5.

Methylprednisolone (MPL) and methylprednisone (MPN) concentration-time profiles in plasma and tissues during 0.3, 3, and 10 mg/h per kg i.v. infusions of MPL. Measured MPL concentrations in low-, medium-, and high-speed infusions are indicated by circles in pink, orange, and red; those for MPN are indicated by triangles in gray, green, and blue. Curves in corresponding colors depict PBPK model fittings of MPL (solid) and MPN (dashed).

Fig. 6.

Methylprednisolone (MPL) and methylprednisone (MPN) concentration-time profiles in plasma and tissues during 0.3, 3, and 10 mg/h per kg i.v. infusions of MPN. Symbols and lines are as defined in Fig. 5.

All parameters for MPL and/or MPN PK (Tables 2 and 3) were estimated with good reliability, as indicated by 31/39 parameters estimated with CV% <30% and the remaining with CV% <50%. In addition, the variance model parameters σ₁ (additive residual) as well as σ₂ (proportional residual) for overall datasets were 0.17 (CV% = 11.4%) and 0.58 (CV% = 2.6%).

Concentration-Time Profiles of MPL and MPN.

As shown in Fig. 3, the MPL and MPN profiles in plasma and most tissues exhibited similar patterns with four decline phases: a brief initial phase, an apparent linear phase, a transition phase, and a long linear terminal phase. Both administered MPL and formed MPN in tissues peaked at the very first sampling time, indicating that MPL and MPN reach quasi-equilibrium quickly between blood and tissues upon bolus dosing and a rapid conversion of MPL to MPN. In Fig. 4, the concentration of formed MPL peaked in a few minutes according to model-predicted curves and was close to administered MPN in the initial phase and surpassed it at around 7.5 hours, indicating a rapid and appreciable conversion of MPN to MPL. Upon infusion (Figs. 5 and 6), both MPL and MPN in plasma and tissues achieved their steady-state concentrations quickly, varying from 0.5 to 4 hours across different tissues and different infusion rates. The steady-state concentrations of MPL in liver and kidney plots from MPN infusion studies were unexpectedly higher than those of MPN (Fig. 6), whereas the steady-state concentrations of the formed MPN or MPL were invariably lower than those of the administered MPL or MPN in the remaining plots of Figs. 5 and 6.

Tissue Partitioning Parameters of MPL and MPN

The parameter estimates relative to tissue partitioning of MPL and MPN are listed in Table 2. This model was robust enough to render a complete array of B_max and K_p estimates for various tissues. The model-predicted binding equilibrium dissociation constant of MPL shared by most tissues was 1.69 ng/ml (equivalent to 4.51 nM) and comparable to that of prednisolone (3.01 ng/ml = 8.36 nM) from a published PBPK model (Li et al., 2020), indicating similar binding affinity toward their targets despite MPL being regarded as a more potent steroid than prednisolone. The highest binding capacity (B_max) of MPL was observed in liver (322.9 ng/ml) followed in descending order by kidney, heart, intestine, skin, spleen, bone, brain, muscle, and the lowest in adipose (2.74 ng/ml). This B_max range basically overlapped that of prednisolone, 50.8∼690.5 ng/ml. For optimal fittings, it was necessary to set lung as an exception with much higher K_{d_MPL_lung} (207.1 ng/ml) and B_{max_lung} (7408 ng/ml) than other tissues. Lastly, the B_{max_MPL} estimate (219.9 ng/ml) of the remainder compartment suggested that substantial MPL-receptor interactions might exist in the unmeasured tissues, including stomach, pancreas, bladder, reproductive organs, glands, and lymph.

TABLE 2.

Tissue-specific partitioning, saturable binding, and transport parameters of methylprednisolone (MPL) and methylprednisone (MPN) obtained in PBPK model fitting The units for B_max, K_d, and K_p are ng/ml, ng/ml, and 1.

Parameter	Estimate (CV%)		Parameter	Estimate (CV%)
Bmax_MPL,liver	322.9 (10.9)		Bmax_MPN,liver	4.23 (23.9)
Bmax_MPL,lung	7408 (17.6)		Kp_MPN,lung	5.38 (8.3)
Bmax_MPL,kidney	153.1 (13.0)		Bmax_MPN,kidney	6.42 (22.2)
Bmax_MPL,heart	71.8 (12.4)		Kp_MPN,heart	2.25 (8.3)
Bmax_MPL,intestine	42.4 (13.4)		Kp_MPN,intestine	3.10 (12.3)
Bmax_MPL,spleen	37.4 (13.6)		Kp_MPN,spleen	2.18 (8.4)
Bmax_MPL,skin	42.1 (11.7)		Kp_MPN,skin	1.85 (7.8)
Bmax_MPL,muscle	8.39 (17.9)		Kp_MPN,muscle	1.50 (7.7)
Bmax_MPL,adipose	2.74 (32.4)		Kp_MPN,adipose	0.93 (8.2)
Bmax_MPL,bone	36.6 (11.7)		Kp_MPN,bone	0.63 (8.3)
Bmax_MPL,brain	14.6 (19.2)		Kp_MPN,brain	0.15 (9.2)
Bmax_MPL,remainder	219.9 (32.1)		Kp_MPN,remainder	15.7 (4.2)
Kd_MPL,other	Equilibrium dissociation constant of MPL in tissues except for lung			1.69 (16.7)
Kd_MPL,lung	Equilibrium dissociation constant of MPL in lung			207.1 (24.5)
Kd_MPN	Equilibrium dissociation constant of MPN in liver and kidney			0.15 (36.1)
CLup_MPL,brain	Bidirectional passive diffusion clearance of MPL in brain			1.02 (45.1)
CLeff_MPL,brain	Unidirectional efflux clearance of MPL in brain			16.66 (40.8)

The tissue/plasma partition coefficients of MPN, K_{p_MPN}, were constant but variable among tissues ranging from 0.15 to 5.38, whereas it exhibited nonlinearity in liver and kidney with B_{max_MPN} of 4.23 and 6.45 ng/ml, indicating lesser binding capacity in tissues than its parent drug, MPL. The equilibrium dissociation constant of 11β-HSD1 for MPN (K_{d_MPN}) was 0.15 ng/ml (equivalent to 0.4 nM). This is one order of magnitude higher than the reported K_d of 11β-HSD1 for its selective and potent inhibitor, SPI-62 (0.035 nM in humans) (Wu et al., 2021), but much lower than that of 11β-HSD1 for its endogenous weak substrates, cortisone and 11-dehydrocorticosterone (2–40 μM for human and mouse) (Chapman et al., 2013).

Low brain MPL concentrations in contrast to the expected high permeability (logP = 1.52) were well captured by incorporating a presumed unidirectional active efflux process (CL_{eff_MPL,brain} = 16.66 ml/h) and a bidirectional passive diffusion process (CL_{up_MPL,brain} = 1.02 ml/h). The estimated CL_eff-to-CL_up ratio of MPL was 16.33 and comparable to that of prednisolone (35.23), indicating that the limited brain distribution of GC is largely due to the active efflux process.

Reversible Conversion and Irreversible Elimination of MPL and MPN

The model-estimated intrinsic clearances for conversion and elimination processes are listed in Table 3, and their calculated systemic or summed clearances are shown in Supplemental Table 1.

The MPL→MPN conversion was predominately in kidney, with intrinsic clearance of 558.7 ml/h and supplemented by nonlinear lung and intestine conversion whose clearances averaged at 74.6 and 51.6 ml/h when the unbound MPL concentration equals to K_{m_LN}, 1319 ng/ml. It was also clearly seen in Fig. 3 that the concentrations of formed MPN in kidney, lung, and intestine were much closer to MPL concentrations than those in other tissues in the initial phase. For the opposite MPN→MPL direction, the intrinsic conversion clearances were 2051 ml/h in liver and 2000 ml/h in kidney. The profiles of MPL and MPN in liver substantially differed upon MPL dosing (Figs. 3 and 5) and approached each other closely upon MPN dosing (Fig. 6). This demonstrated remarkable hepatic MPN→MPL conversion, whereas this phenomenon is far less in kidney due to the coexistence of MPL/MPN conversion in both directions. In comparison, the total systemic back conversion process was more than three times greater than that of MPL→MPN conversion, 1342.4 ml/h (CL_NL) versus 429.3 ml/h (CL_LN). Based on systemic clearances, the irreversible hepatic metabolism, pulmonary metabolism, and renal excretion accounted for 44%, 35%, and 21% of the total MPN elimination (CL_MPNe = 2758.2 ml/h), whereas total MPL elimination clearance (CL_MPLe = 789.5 ml/h) was lower and attributed to hepatic metabolism.

Taken together, the metabolic processes operating on the metabolite, MPN, were greater than those on parent, MPL, with irreversible conversion of metabolite being the dominant kinetic term. In addition, the exposure enhancement (EE) and recycled fraction (RF) of parent MPL in rats was 1.13 and 0.12 (Supplemental Table 1), indicating that approximately 12% of dosed MPL was involved in metabolic recycling, and thus its systemic exposure is equivalent to giving 113% of the original dosage in the absence of recycling. The EE and RF values of the MPL/MPN pair in rats were smaller than those in rabbits, 1.42 and 0.29, and smaller than those of cortisol and prednisolone pairs observed in humans (Ebling and Jusko, 1986).

Discussion

Blood Partitioning

In plasma, MPL was reported to bind to albumin in a nonsaturable, low-affinity manner at 60.5%∼63.1% in rats (Haughey and Jusko, 1991; Ayyar et al., 2019). There is also little difference of plasma protein binding between MPL and MPN in other species, human and rabbit (Ebling et al., 1986). Judging from Supplemental Fig. 1, the MPL and MPN concentrations in red blood cells (RBCs) calculated according to C_RBC = [C_blood − C_plasma·(1 − HCT)]/HCT, where HCT is the hematocrit of male Wistar rat, 0.45 (Kampfmann et al., 2012), was similar to their unbound concentrations in plasma, indicating rapid equilibriums between plasma and RBCs and absence of binding in blood cells. Accordingly, the blood partitioning, R_b, was calculated using R_b = f_u,plasma·HCT + (1 − HCT) (Song et al., 2020), in which f_u,plasma is the free fraction of MPL or MPN in plasma, 0.395 (Ayyar et al., 2019), and yielded 0.73. This is comparable to the measured mean R_{b_MPL} (0.64) and R_{b_MPN} (0.63) and suggested limited entry of MPL and MPN into RBCs.

Saturable Tissue Binding

The well known target of MPL, GR, is ubiquitously expressed intracellularly in almost all tissues and is the theoretical basis of the B_{max_MPL} assignment in our model. There are two additional potential binding targets of MPL, mineralocorticoid receptors (MRs) and 11β-HSD2. The MR binds GC and aldosterone with similar affinity in vitro, whereas MR only binds aldosterone in vivo. The colocation of MR and 11β-HSD2, which catalyzes the rapid inactivation of GC to inert keto form thus only allows the authentic ligand aldosterone to access MR (Chapman et al., 2013). The sites of the nonlinear MPN binding were assumed to coincide with that of MPN→MPL conversion, considering that 11β-HSD1 is a unique known target of MPN. The nonlinear binding sites of MPL and MPN were preliminarily assessed according to the correlation curves of tissue versus plasma concentrations of MPL and MPN, as shown in Supplemental Fig. 2. The correlation curve is sigmoidal shifting toward the right as plasma concentrations increase when nonlinear binding of MPL or MPN exists, or else it will display as a straight line (Supplemental Fig. 3).

Our model rendered a complete array of MPL tissue binding capacities (B_{max_MPL}). Not surprisingly, liver topped the list (322.9 ng/ml), owing to the most abundant expression of GR in rat liver (Ballard et al., 1974), whereas kidney was second (153.1 ng/ml), probably owing to its coexpression of additional binding targets. Besides kidney, the distribution of MR and 11β-HSD2 are confined to other classic aldosterone targets such as colon, salivary glands, and brain hippocampus, which explains the moderate B_{max_MPL} in intestine (42.4 ng/ml). The comparison of GR density among skeletal muscle, adipose, and lung in rats in their baseline GR mRNA expressions was approximately 1: 2.1: 2.7 (Ayyar et al., 2017), indicating the relative lower expression of GR in muscle that agrees with its B_{max_MPL,muscle} of 8.39 ng/ml. Even though the affinities of MPL to MR and 11β-HSD2 are supposed to be greater than to GR, supported by MR having 10-fold higher affinity for cortisol than GR (Chapman et al., 2013) and that 11β-HSD2 captures and inactivates GC efficiently before they access MR, an unique K_{d_MPL} was found in our model and estimated as 1.69 ng/ml (equivalent to 4.51 nM).

Tissue Partition Coefficients

According to K_p estimates of MPN in Table 2, tissues could be roughly classified into four tiers: first-tier with high or plasma concentration–sensitive K_p (liver, lung, kidney); second-tier with moderate K_p (heart, intestine, spleen, skin, muscle); third-tier with modest K_p (adipose and bone); and fourth-tier with low K_p (brain, owing to BBB). This tissue hierarchy holds true for other steroids such as dexamethasone (Song et al., 2020) and prednisolone (Li et al., 2020). As for the nonlinear K_p of MPL and MPN, the calculations based on predicted areas under the curve (AUCs) and steady-state concentration ratios are shown in Supplemental Tables 2 and 3. The K_p decreases in tissues involving saturable binding as plasma concentrations increase.

Conversion and Elimination

An extensive literature documents the role of 11β-HSDs in systemic turnover of endogenous steroids and exogenous GCs as reviewed in Chapman et al. (2013). 11β-HSD1, the enzyme that predominately catalyzes the conversion of inactive GC back to active forms, is ubiquitously and heterogeneously expressed in most tissues and hence modulates tissue-specific reactivation of nonactive keto GC, whereas the MPN→MPL conversion was incorporated and functioned well in liver and kidney in present PBPK model. The highest expression of 11β-HSD1 in liver (Tannin et al., 1991) is consistent with an intrinsic MPN→MPL conversion clearance of 2051 ml/h. This enzyme is also expressed in rat kidney (Krozowski, 1999), and the MPN→MPL conversion was 2000 ml/h and comparable to that in liver. The 11β-HSD2 determines not only saturable tissue binding of MPL but the MPL→MPN conversion. The conversion clearance in kidney was linear and accounted for ∼71% of total CL_LN and substantiates the highest expression of 11β-HSD2 in kidney (primarily in proximal tubule) (Krozowski, 1999). The MPL→MPN conversion in intestine is nonlinear and in good agreement with the well recognized 11β-HSD2 enzyme distribution in colon (Zhou et al., 1995).

Other than reversible metabolism, MPL is metabolized irreversibly by cytochromes P450 (P450s) 3A enzymes (Czock et al., 2005), and the systemic hepatic clearance was estimated to be 789.5 ml/h in rats. However, renal excretion of MPL was not included, as only 0.5% unchanged MPL was detected in rat urine (Kong and Jusko, 1991). MPN was metabolized further in rat liver with clearance of 1199.1 ml/h and also excreted in kidney with clearance of 582.5 ml/h, which is evidenced as both MPN itself and its phase I and phase II metabolites were identified in human urine after MPL dosing (Panusa et al., 2011; Matabosch et al., 2013).

The terminal phases of blood and tissue profiles of MPL in Fig. 3 were relatively long, as MPL is retained by its widespread binding targets and the parallel MPN terminal decline can be attributed to the interconversion equilibrium with MPL. We also found that the terminal phase of MPL and MPN profiles stayed the same no matter which form of steroid was administered, as shown in Supplemental Fig. 4.

Special Disposition in Lung

The lung is a distinctive tissue for MPL from both pharmacodynamic and PK points of view. It was technically necessary to set a separate equilibrium dissociation constant of MPL (i.e., K_{d_MPL_lung} = 207.1 ng/ml) for optimized fitting, which was 100-fold higher than the universal one adapted for other tissues. The binding capacity (7408 ng/ml) was also much greater than in other tissues. Together with a much higher K_{p_MPN} than other tissues and the considerable distribution of MPL (and MPN) in lung reported previously (Vichyanond et al., 1989; Greos et al., 1991), a rational basis is offered for the superiority of MPL in ARDS treatment. Our model indicated the presence of MPL→MPN conversion in lung, but the pulmonary expression of 11β-HSD2 is still under debate. The 11β-HSD2–mediated oxidase activity for several GCs was high in rat lung microsomes (Hundertmark et al., 1993), but another study found that the adult rat lung barely expresses 11β-HSD2 (Alikhani-Koopaei et al., 2004). The MPN elimination in lung averaged 73.56 ml/h, which is explained by the existence of metabolizing P450s in lung such as 2A13, 2F1, and 3A5 (Ding and Kaminsky, 2003). Further possible explanation is the inducible expression of certain P450s in lung by MPL or MPN, as 3A5 mRNA was induced 8-fold by dexamethasone in human lung cell lines (Hukkanen et al., 2000).

Model Limitations and Advantages

One limitation of our current model is that the 11β-HSD1–mediated MPN→MPL conversion was modeled as linear enzymatic reaction supplemented with nonlinear MPN-11β-HSD1 binding. A TMDD conception pertaining to 11β-HSD1 might apply, but this cannot be differentiated in our fittings. The unique MPL disposition in lung remains to be explored further. The observations at later time points in Fig. 4 were not captured well; these concentrations were very low and sparse and thus relatively uninformative in modeling and somewhat less important in the overall context of the PBPK model.

This comprehensive whole-body PBPK model was robust and well characterized the disposition of MPL and its metabolite MPN in rats. It included the processes of MPL/MPN blood partitioning, ubiquitous MPL saturable tissue binding, nonlinear and linear MPN tissue partitioning, BBB-associated MPL diffusion and efflux, irreversible elimination of MPL and MPN, and reversible conversion between MPL and MPN. This model was developed not only based on rat-sourced data and reports but also refers to and confirms information from human studies, such as the distribution of 11β-HSD isozymes and the elimination routes of MPL and MPN. The tissue binding of MPL and MPN and their underlying mechanisms were assumed to be universal across species, whereas there are species differences in the rates of conversion, elimination, and other disposition processes. The mechanisms underlying our PBPK model are thus generally applicable to humans and perhaps other species and provide instructive insights into its clinical pharmacodynamics.

Abbreviations

BBB

blood-brain barrier

C

drug concentration

CL

clearance

GC

glucocorticoid

GR

glucocorticoid receptor

HCT

hematocrit

11β-HSD

11β-hydroxysteroid dehydrogenase

IS

internal standard

K_d

equilibrium dissociation constant

K_m

Michaelis-Menten constant

K_p

tissue/plasma partition coefficient

LC-MS/MS

liquid chromatography–tandem mass spectrometry

MPL

6α-methylprednisolone

MPN

6α-methylprednisone

MR

mineralocorticoid receptor

P450

cytochrome P450

PBPK

physiologically based pharmacokinetics

PK

pharmacokinetics

Q

blood flow

R_b

blood-to-plasma ratio

RBC

red blood cell

TMDD

target-mediated drug disposition

Authorship Contributions

Participated in research design: Yu, Jusko.

Conducted experiments: Yu.

Performed data analysis: Yu, Jusko.

Wrote or contributed to the writing of the manuscript: Yu, Jusko.