Preclinical Modeling and Simulation to Explore the Tissue/Plasma Exposure and Pharmacodynamic Effect of Vildagliptin in Diabetes Treatment

ABSTRACT

Vildagliptin (VDG) is a dipeptidyl‐peptidase‐4 (DPP‐4) inhibitor used for type 2 diabetes (T2DM) treatment. Viewing to improve VDG treatment, a population pharmacokinetic (popPK) model was built to describe drug plasma, free liver and muscle concentrations determined by microdialysis in healthy and diabetic animals following 50 mg/kg i.v. bolus administration. A four‐compartment popPK model with linear elimination and bidirectional transport between tissues and the central compartment described the data with diabetes as a covariate in Q₁ and Q_out,liver. The pharmacokinetic parameters of VDG were scaled to humans using allometry, and used to simulate VDG tissue concentrations in patients with T2DM and relate them with the DPP‐4 inhibition by an I _max model. The efficacy of VDG was evaluated considering 80% and 92% DP‐IV inhibition during the entire dosing interval. VDG 100 mg q24 h achieved 80% DPP‐4 inhibition in plasma, but not in tissues. Although q12 h dosing interval reached 80% enzyme inhibition in plasma for > 25 mg doses, only the 100 mg reached this goal in muscle. The 92% enzyme inhibition was achieved in plasma for 50 and 100 mg q12 h but none of the dose regimens investigated reached this inhibition in tissues.

Study Highlights.

• What is the current knowledge on the topic?

  • ○
    Vildagliptin (VDG) is a dipeptidyl peptidase IV inhibitor approved to treat type 2 diabetes. Some studies reported that VDG plasma concentrations are similar in healthy volunteers and diabetic patients. Furthermore, a previous preclinical investigation from our group showed that VDG free concentration in muscle and liver in rats with diabetes does not differ significantly.
• What question did this study address?

  • ○
    Are VDG free plasma concentrations a good surrogate for the pharmacologically active concentrations at relevant tissues? Does diabetes alter VDG free tissue concentrations? Do these possible changes in free tissue concentration impact the effect of the drug, requiring a dosage adjustment?
• What does this study add to our knowledge?

  • ○
    In the present study we demonstrate that VDG free plasma concentrations are not reliable surrogates for free concentrations in liver and muscle, especially in diabetic conditions. With simulations we demonstrated that free VDG exposure in these tissues generates DPP‐IV inhibition higher and for the longest time when the doses were administered twice daily.
• How might this change drug discovery, development, and/or therapeutics?

  • ○
    This study underscores the importance of evaluating tissue‐specific drug distribution and the impact of diseases on PK profiles and PK/PD, guiding the optimization of dosing regimens for a DPP‐4 inhibitor, ensuring therapeutic concentrations are achieved in relevant tissues for maximal clinical efficacy.

1. Introduction

Diabetes mellitus is a chronic and complex metabolic disorder characterized by increased glucose levels in the blood, due to a lack or dysfunction of the insulin hormone. Type 2 diabetes mellitus (T2DM) occurs when insulin secretion is lower than that required to maintain glucose blood concentration, or when, despite normal insulin secretion, there are hormone‐resistant cells [1, 2].

Single‐drug pharmacological treatment for T2DM is associated with significant side effects and tolerance, leading to the use of drugs' association to improve glucose blood levels maintenance [3]. The development of new therapies and the optimization of existing treatments that promote glycemic control with more constant plasma profiles resulting in fewer adverse effects and better patients' outcomes have been of great interest to researchers and clinicians [2, 3, 4].

Vildagliptin (VDG) is a dipeptidyl peptidase IV (DPP‐IV) inhibitor that blocks the enzyme degradation of incretins, increasing glucose‐dependent insulinotropic polypeptide (GIP) and glucagon‐like peptide (GLP)‐1 action [5, 6, 7]. Vildagliptin shows high oral bioavailability (85%–90%), reaching peak plasma concentrations in 1–2 h after dosing, and plasma protein binding of less than 10%. After oral administration to monkeys, VDG presented a volume of distribution of 0.7 L/kg and a clearance of 1.5 L/h/kg [7]. Vildagliptin is 55% catabolized by non‐CYP‐mediated hydrolysis, mainly in the liver, forming an inactive metabolite; 1.6% of the dose is metabolized by CYP enzymes oxidation and glucuronidation. The renal pathway is responsible for 85% of VDG and its metabolite elimination, being 21% of the dose eliminated as unchanged drug. The remainder of the drug dose is eliminated in the feces. The pharmacokinetics of VDG have been evaluated in healthy volunteers [8, 9] and in T2DM patients [10] showing that the disease does not modify the drug's plasma pharmacokinetic parameters [11, 12].

Gene expression studies show a high expression and activity of the DPP‐IV enzyme in tissues like the lung, small intestine, muscle, kidney, liver, and pancreas [13]. However, there are no reports in the literature measuring VDG concentration in these tissues in humans. It is well established that free drug concentrations at the site of action are responsible for the pharmacological effect [14, 15]. However, accessing free tissue concentrations in the clinical setting can be tricky, and plasma concentrations, an easily accessible sampling site, may not be a good surrogate for tissue concentrations [16, 17]. Previous reports from alloxan‐induced diabetic rats described VDG free concentrations in muscle and liver, demonstrating that the drug exposure in both tissues is similar [18]. Still, the impact of diabetes on VDG tissue exposure and the relationship between plasma and free tissue concentrations need to be elucidated contributing to improving this drug's clinical use to treat T2DM.

Therefore, this work aimed to develop a population pharmacokinetic model (popPK) to describe VDG plasma, free liver, and free muscle concentrations observed in an animal model of diabetes. The model was then applied to simulate free plasma and tissue concentrations expected in diabetic patients, and correlate them with DPP‐IV inhibition viewing to predict the efficacy of VDG current T2DM treatment.

2. Methods

2.1. Chemicals and Reagents

Vildagliptin was obtained from Ontario Chemicals (Ontario, Canada). Cimetidine was purchased from Changzhou (China). Ammonium acetate and sodium hydroxide were obtained from QM Reagentes (Barueri, Brazil). HPLC grade acetonitrile and methanol were purchased from Tedia (Fairfield, Ohio, USA). HPLC water from Millipore's Milli‐Q System was used in the analysis. Sodium chloride, potassium chloride, calcium chloride, and all other chemicals and reagents were of analytical grade and purchased from commercial sources. Ringer's solution consisted of 147 mM of NaCl, 1.3 mM of CaCl2, and 4 mM of KCl.

2.2. Bioanalytical Method for VDG Quantification

Vildagliptin quantification in plasma and microdialysate samples was conducted using a validated LC–MS/MS method previously described [18]. For analytical analysis details, see Supporting Information.

2.3. Pharmacokinetic Experimental Design

This study was approved by the Ethics Committee in Animal Use from the Federal University of Rio Grande do Sul (CEUA/UFRGS #20352) and was conducted in compliance with the principles of laboratory animal care from the National Council for Animal Experimentation (CONCEA/Brazil). Male Wistar rats (250–350 g, 8 weeks old) were obtained from the University's Center for Laboratory Animals Reproduction and Experimentation (CREAL, Porto Alegre, Brazil). All animals were kept in a controlled environment (22°C ± 2°C, 65% humidity, in a 12 h light/dark cycle) with free access to standard rodent chow and filtered water.

2.3.1. Animal Model of Diabetic

Diabetes‐induced protocol was previously validated in the laboratory using alloxan as a chemical inducer according to the protocol described by Lerco et al. and Carvalho et al. [19, 20]. Briefly, an aqueous solution of alloxan 2% at a dose of 40 mg/kg was administered intravenously (i.v.) to the animals. Seventy‐two hours after induction a blood glucose test was performed with a portable glycosometer, and the animals were considered diabetic when glucose > 300 mg/dL. The pharmacokinetic experiments were performed 4 weeks after induction, when the animals presented the metabolic alterations of diabetes [19, 20].

2.3.2. Plasma Pharmacokinetics

Animals were randomly distributed into two groups to evaluate VDG plasma pharmacokinetic after a 50 mg/kg i.v. bolus dose to healthy (n = 7) and diabetic (n = 5) rats. Animals were anesthetized with ethyl carbamate (1.25 mg/kg, intraperitoneally) and the carotid artery was cannulated for blood sampling at predetermined times: 0.083, 0.166, 0.2, 0.5, 0.75, 1, 1.5, 2, 4 and 6 h after dosing. Plasma (100 μL) was obtained from heparinized blood supernatant after centrifugation (20,000 × g, 4°C, 10 min) and frozen until analysis.

2.3.3. Muscle and Liver Pharmacokinetics

Free VDG concentrations were assessed in the muscle and liver of healthy (n = 7 and 6, respectively) and diabetic (n = 6 and 6, respectively) animals by microdialysis in anesthetized (urethane, 1.25 g/kg i.p.) rats. A microdialysis probe (4 mm, cut‐off 20 KDa, CMA Sweden) was inserted in the rat's muscle or liver and Ringer solution was perfused into the probes at a flow rate of 1.0 μL/min. Animals received vildagliptin 50 mg/kg i.v. bolus through the femoral vein and microdialysate samples (40 μL) were collected every 40 min up to 6 h and frozen until analysis.

2.3.4. Microdialysis Probe Calibration

Previous in vitro studies have indicated that VDG does not bind to probe and microdialysis connecting tubes, confirming that the drug recovery is concentration independent [18]. To assure the correct measurement of VDG concentrations in tissues, microdialysis probe calibration was performed in other animals than the pharmacokinetics analysis, by retrodialysis [21] in vivo (liver and muscle) in healthy and diabetic animals (n = 3/group). After probe insertion in liver or muscle, and 1 h stabilization using Ringer as perfusion fluid, the perfusion fluid was switched to VDG 500 ng/mL in Ringer at a flow rate of 1.0 μL/min. After another hour of stabilization, four 40 min microdialysate samples were collected and drug quantified. Relatively recovery (RR%) was calculated by Equation (1):

$$\mathrm{RR}\left(\%\right)=100-\left(\frac{C_{\text{dial}}}{C_{\text{perf}}}\times 100\right)$$ (1)

where $C_{\text{dial}}$ is the VDG concentration quantified in the microdialysate and $C_{\text{perf}}$ is the concentration of the perfusate solution. VDG free tissue concentrations corrected using the in vivo RR% was used to build the model.

2.4. Population Pharmacokinetic Modeling

2.4.1. Softwares

Plasma and tissue data were log‐transformed and analyzed by population pharmacokinetic (popPK) approach using a nonlinear mixed effects model with the software NONMEM version 7.4 (Icon Development Solutions, Ellicott City, MD, USA). All estimations applied the first‐order conditional estimation method with interaction (FOCE‐I). PsN toolkit (Perl‐speaks‐NONMEM) version 4.9.0 was used for complementary analysis. Model management was done in PIRANA version 2.9.9 (Pirana Software and Consulting). Ggplot2 and xpose4 libraries for R version 4.1.1 and RStudio, version 1.4.1717 (The R Foundation for Statistical Computing) were used for graphical analysis.

2.4.2. Model Building

For the model building plasma free concentrations were estimated considering rat's plasma protein binding of 9% [9]. Free tissue concentrations obtained by microdialysis were described by the integral over each collection interval [22]. The structural model was performed sequentially, first establishing the structural model for plasma data and next including the tissue data. The model was parameterized by clearance (CL), volume of distribution (V) and intercompartmental clearances (Q), using differential equations (ADVAN 13 TRANS1). Inter‐individual variability (IIV) was modeled exponentially, and residual variability was described separately for plasma and microdialysate data (muscle and liver) with a log‐additive error model.

2.4.3. Model Selection

Model selection was guided by (i) a significantly change in the value of the objective function (OFV), considering a decrease of 3.84, or 6.64 points in OFV (p < 0.05 or p < 0.01, respectively) for the evaluation of two models; (ii) visual exploration of goodness‐of‐fit (GOF) plots, and (iii) precision of model parameters reflected as the relative standard error (RSE%) computed as the ratio between the standard error and the parameter estimate.

2.4.4. Covariate Analysis

Diabetes was evaluated as a covariate in model parameters, using a categorical variable. Status was defined as 0 for healthy and 1 for diabetic animals. Additive and proportional models were tested. Covariates were selected according to forward inclusion (p < 0.05) and backward elimination (p < 0.01), considering the conditions for model selection, visual exploration of GOF plots, precision of model parameters and the reduction of more than 5% in IIV values.

2.4.5. Model Evaluation

Evaluation of the selected popPK model was performed by visualization of GOF plots and visual predictive checks (VPC), observation of RSE%, and conditional number. Model stability was checked with a nonparametric bootstrap analysis, performing 1000 resampling from the final model with the median and confidence interval (CI) of the 5th and 95th quartiles of each parameter.

2.5. Allometry

For estimating human parameters, we used an allometric equation based on weight and allometric exponents [23]. The parameters estimated in the popPK model developed with VDG plasma and tissue concentrations in rats were allometrically scaled for humans using Equation (2):

$${\text{Parameter}}_{\text{human}}={\text{Parameter}}_{\mathrm{rat}}\times {\left(\frac{{\text{Weight}}_{\text{human}}}{{\text{Weight}}_{\mathrm{rat}}}\right)}^b$$ (2)

where b is defined as the allometric exponent, being 0.75 for clearances parameters and 1 for volumes.

Parameters regarding absorption, such as absorption constant (Ka) and absolute bioavailability (F _abs), used in the simulations were reported by Landersdorfer et al. [24] for VDG plasma concentrations following VDG oral dosing of 10, 25, and 100 mg to diabetic patients, being F _abs of 77.2% and absorption rate constants (K _a1) = 1.26 h⁻¹ and K _a2 = 1.05 h⁻¹.

2.6. Pharmacokinetic/Pharmacodynamic Model

Using the concentrations for patients in plasma and tissue estimated by allometry, a pharmacokinetic/pharmacodynamic (PK/PD) relationship was established to predict this DPP‐4 effect in diabetic patients at a tissue level.

The dose regimens tested were 25, 50, and 100 mg once and twice daily, in 1000 simulations of steady‐state concentrations. The free concentrations simulated for humans in liver and in muscle were used to estimate the percentage of DPP‐4 inhibition with the maximum inhibition model, calculated as Equation (3):

$$\mathrm{DPP}-4\text{inhibition}=\frac{I_{\max }\times C_{\mathrm{u}}}{C_{\mathrm{u}}+{\mathrm{IC}}_{50}}$$ (3)

where I _max is the maximum enzyme inhibition, IC₅₀ is the VDG concentration where there is 50% of DPP‐4 inhibition, set as 1350 mg/L [10, 25] (4.5 nmol/L); and C _u is the free simulated concentrations of VDG in muscle and liver. Clinical efficacy was evaluated by the percentage of DPP‐IV inhibition during the entire duration of dosing interval, considering the steady‐state concentrations expected on the 5th day of treatment. An efficacious treatment should result in at least 80% enzyme inhibition for the whole duration (100%) of the dosing interval to adequately maintain blood glucose concentration [26]. Furthermore, a 92% of enzyme inhibition results in adequate reduction of glycosylated hemoglobin (HbA1) level [27].

3. Results

Microdialysis probes relative recovery determined in vivo in the liver and muscle of healthy and diabetic rats was 31.2% ± 5.0%, similar to the value previously determined in vivo in the liver of rats with diabetes by de Andrade et al. [18], showing that recovery was independent of the tissue or animal health condition investigated. Total plasma, free muscle, and free liver VDG concentrations for healthy and diabetic rats are shown in Figure 1. A total of 254 observations from plasma, free muscle, and free liver experiments in healthy and diabetic conditions were used for the model building. Table S1 describes the number of animals and observations in each group. A visual exploration of Figure 1 demonstrates that VDG free concentrations in the liver of healthy rats are smaller than those observed in rats with diabetes.

FIGURE 1.

Vildagliptin concentrations versus time profiles in total plasma (n = 7 and 5), free muscle (n = 6 and 6), and free liver (n = 7 and 6) of healthy rats and rats with diabetes, respectively. Mean ±SD.

VDG plasma concentrations were described by a two‐compartment model with linear elimination, parameterized by total clearance (CL), volume of the central compartment (V ₁), intercompartmental clearance from the central to the peripheral compartment (Q ₁) and volume of the peripheral compartment (V ₂). The inclusion of muscle and liver VDG free concentrations was tested using different models, and the best fit was obtained with the inclusion of the tissues as a third and fourth compartment, with the volumes of tissue compartments (V ₃ and V ₄, respectively) and different bidirectional transport from plasma to tissues (Q _in,muscle and Q _in,liver) and the return from tissues to plasma (Q _out,muscle and Q _out,liver). Figure 2 shows the structural model for VDG, describing all compartments and parameters. The differential equations used to estimate concentration in each compartment are (Equations (4), (5), (6), (7)):

$$\begin{array}{c}\frac{\text{dCentral}}{\mathrm{dt}}=\frac{Q_1}{V_2}\times A_2+\frac{Q_{\text{out},\text{muscle}}}{V_3}\times A_3+\frac{Q_{\text{out},\text{liver}}}{V_4}\times A_4\hfill \\ -\left(\frac{\mathrm{CL}}{V_1}+\frac{Q_1}{V_1}+\frac{Q_{\text{in},\text{muscle}}}{V_1}+\frac{Q_{\text{in},\text{liver}}}{V_1}\right)\times A_1\hfill \end{array}$$ (4)

$$\frac{\text{dPeripheral}}{\mathrm{dt}}=\frac{Q_1}{V_1}\times A_1-\frac{Q_1}{V_2}\times A_2$$ (5)

$$\frac{\text{dMuscle}}{\mathrm{dt}}=\frac{Q_{\text{in},\text{muscle}}}{V_1}\times A_1-\frac{Q_{\text{out},\text{muscle}}}{V_3}\times A_3$$ (6)

$$\frac{\text{dLiver}}{\mathrm{dt}}=\frac{Q_{\text{in},\text{liver}}}{V_1}\times A_1-\frac{Q_{\text{out},\text{liver}}}{V_4}\times A_4$$ (7)

FIGURE 2.

Structural PK/PD model developed for VDG free plasma, muscle and liver concentrations and their DPP‐4 inhibition. Figure shows the schematic representation of the 4‐compartment model describing plasma and microdialysate samples, and the relationship of these free concentrations and the DPP‐4 inhibition. A ₁, A ₂, A ₃, and A ₄ are the amounts of drug in central, peripheral, muscle and liver compartments, respectively; V ₁, V ₂, V ₃, and V ₄ are the volumes of distribution in central, peripheral, muscle, and liver compartments, respectively; CL is the central clearance, and Q ₁ the intercompartmental clearance between central and peripheral compartments; Q _in and Q _out for muscle and liver are the intercompartmental clearances of entrance and exit in each tissue compartment, respectively; C _u is the simulated VDG free tissue or plasma concentration in humans; I _max is the maximum inhibition of DPP‐4, and IC₅₀ is the VDG concentration that achieves 50% of DPP‐4 inhibition.

Interindividual variability (IIV) included exponentially was estimated for CL, Q ₁, V ₃, Q _out,liver and V ₄. The induced diabetes was responsible for significantly reducing the OFV and the variability in Q ₁ and Q _out,liver. The estimation of these two parameters separately for diabetic and healthy animals improved model stability.

Observations versus individual or population predicted values displayed good prediction of the model, corroborated by the goodness of fit plots in Figure 3, which does not present bias in the predictions. The model was successfully evaluated by VPC analysis, graphically disposed in Figure 4, where the observations are within the predicted interval. Table 1 shows the estimated model parameters for the popPK model, with residual standard error (RSE %), shrinkage, and bootstrap analysis. The bootstrap resampling had 96% of successful runs confirming the model stability.

FIGURE 3.

Goodness of fit plots for the final model. Blue, red, and green dots represent free plasma, muscle and liver data, respectively.

FIGURE 4.

Visual predictive checks (VPCs) for the final model stratified by plasma, muscle, and liver in healthy and diabetic rats. VPCs are based on 1000 simulations, where dots are the observations and dashed lines and shadow areas are the 10th, 50th, and 90th percentiles.

TABLE 1.

Parameters estimated for vildagliptin popPK model in healthy rats and rats with diabetes.

Parameters	Estimate (% RSE)	IIV (% RSE)	Shrinkage %	Nonparametric bootstrap median (95% CI)
CL (L/h/kg)	2.72 (5)	19.2 (19)	16	2.64 (2.34–2.92)
V 1 (L/kg)	1.11 (9)	—	—	1.09 (0.85–1.29)
Q 1,healthy (L/h/kg)	0.338 (21)	65.4 (20)	30	0.322 (0.217–0.428)
Q 1,diabetic (L/h/kg)	2.61 (21)	65.4 (20)	30
V 2 (L/kg)	1.85 (12)	—	—	1.54 (0.73–1.98)
Q in,muscle (L/h/kg)	2.09 (22)	—	—	1.92 (1.19–3.17)
Q out,muscle (L/h/kg)	14.3 (23)	—	—	13.51 (8.22–23.27)
V 3 (L/kg)	2.55 (26)	76.6 (29)	30	2.47 (1.29–4.12)
Q in,liver (L/h/kg)	2.5 (31)	—	—	1.74 (0.62–3.87)
Q out,liver,healthy (L/h/kg)	218 (53)	72.0 (21)	40	165 (49–419)
Q out,liver,diabetic (L/h/kg)	19.7 (49)	72.0 (21)	40
V 4 (L/kg)	0.856 (49)	181.4 (26)	43	2.24 (0.52–6.69)
Residual error (ng/L)
Plasma	0.213 (21)	—	—	0.209 (0.148–0.267)
Muscle	0.217 (10)	—	—	0.208 (0.158–0.251)
Liver	0.159 (15)	—	—	0.160 (0.128–0.206)

Note: IIV, inter‐individual variability expressed as CV (%) calculated as ω × 100, where ω, is the standard deviation of the variance of the random effect; RSE (%): Relative standard error. Bootstrap 1000 times, 96% successful runs. All other definitions are in the text.

Using the PK parameters estimated from the popPK model (Table 1) and scaled to humans, we were able to predict free tissue concentrations for patients with T2DM assuming that diffusion is the main distribution mechanism between plasma and tissues (liver and muscle). Parameters for patients are described in Table S2. Simulated steady‐state free plasma and tissue concentration‐time profiles (96–120 h) for different oral dosing regimens given to patients with T2DM are shown in Figure 5 in the right panels. All dose regimens investigated (oral 25, 50, and 100 mg once and twice daily) reached free plasma concentrations above the IC₅₀. However, 25 and 50 mg in both tissues once‐daily dose administration produced plasma concentrations under the IC₅₀ after 16 and 20 h, respectively. As expected, the twice‐oral daily regimens produced less fluctuation, ensuring free tissue concentrations above the IC₅₀ for the three doses tested during the whole dosing interval.

FIGURE 5.

Simulated vildagliptin concentrations and DPP‐IV inhibition. Vildagliptin concentration‐time profiles at steady state after 25, 50, and 100 mg after once‐daily (right panel, top) and twice daily (right panel, bottom). DPP‐4 inhibition in time after 25, 50, and 100 mg once daily (left panel, top) and twice daily (left panel, bottom). Dotted line represents the IC₅₀ = 1.35 ng/mL (left panels), and 80% inhibition (right panels).

Simulated free concentrations at steady‐state were used to predict the DPP‐4 inhibition (Figure 5, left panels), according to the structural model described in Figure 1. The time spans (in percentage) where the inhibition of DPP‐IV remains above 80% or 92% considering the expected free VDG concentrations in each compartment following the different dose regimens investigated, determined at steady‐state for the fifth day of treatment, are shown in Table 2.

TABLE 2.

Percentage of time in a 24‐h interval at steady state where DPP‐IV inhibition is above 80% and 92% after simulated VDG doses of 25, 50, and 100 mg once (q24 h) or twice daily (q12 h).

Compartment	Dose (mg)	Percentage of dosing interval with indicated enzyme inhibition
		80% of inhibition		92% of inhibition
		q24 h	q12 h	q24 h	q12 h
Plasma	25	65.31	100.00	32.65	77.55
	50	95.92	100.00	51.02	100.00
	100	100.00	100.00	77.55	100.00
Muscle	25	16.33	36.73	0.00	0.00
	50	30.61	69.39	8.16	20.41
	100	48.98	100.00	22.45	53.06
Liver	25	16.33	44.90	0.00	0.00
	50	32.65	85.71	8.16	24.49
	100	57.14	100.00	24.49	65.31

Note: Orange shadow areas represent the respective treatment which presented 100% of enzyme inhibition.

Free plasma concentrations generated DPP‐4 inhibition greater than 80% for both time intervals (q24h and q12h) following VDG 100 mg oral dosing. However, the free tissue concentrations after once‐daily doses did not produce relevant DPP‐4 inhibition when the dose is given once a day. Twice‐daily regimen resulted in 80% plasma DPP‐IV inhibition for 25, 50, and 100 mg in plasma. Only with 100 mg in free muscle concentrations warrants 100% enzyme inhibition. However, for the strict cut‐off of 92% enzyme inhibition, only plasma concentrations following 50 and 100 mg q12 h reached levels that guarantee 100% of the dosing interval inhibition.

4. Discussion

Vildagliptin, a DPP‐4 inhibitor approved for T2DM, already has its plasma pharmacokinetic profile in volunteers and patients described [9, 25, 28]. However, free VDG concentrations in tissues, which are important to determine body exposure to the drug and its pharmacological effect, have not been evaluated in humans. In this work, a preclinical study was used to build a popPK model that was applied to predict VDG free concentration in human tissues. The predicted free tissue concentrations were used as the input in a maximum inhibition model to estimate DPP‐IV inhibition for different dose regimens used in the clinic.

DPP‐4 expression and activity is major in tissues like lung, small intestine, muscle, kidney, liver, and pancreas [13]. The function of DPP‐4 mediating local inflammation and insulin resistance is concentrated in adipose and hepatic tissues. Therefore, the therapeutic effect produced by DPP‐4 inhibitors takes place in these tissues [15, 29]. Using microdialysis, we demonstrated that free plasma concentrations of VDG are poor predictors of its unbound concentrations in muscle and liver tissues. Furthermore, diabetes does not influence plasma levels but significantly alters the tissue distribution of the drug.

The VDG popPK model was initially built using plasma concentrations following i.v. dosing, which were described using two compartments, similar to other models reported for VDG [24], with a total CL estimated as 2.72 L/h/kg, V ₁ 1.1 L/kg, Q ₁ 0.338 L/h/kg, and V ₂ 1.85 L/kg. The only study found in the literature about vildagliptin pharmacokinetics in rats reported a CL of 2.9 L/h/kg [30].

Two extra compartments were linked to the central compartment by bi‐directional intercompartmental clearances to describe free liver and free skeletal muscle drug concentrations. The bi‐directional intercompartmental clearances were necessary to account for the differences in the rates of VDG distribution to and from the tissues. The use of bi‐directional clearances has already been reported for other drugs to describe lung, cerebral spinal fluid and soft tissues concentrations [31, 32, 33]. The PopPK model developed does not reproduce physiological blood flows or tissue volumes. For example, the physiological interstitial volumes in a 280 g rat are approximately 15.8 mL for muscle and 2.56 mL for liver [34]. In contrast, our PopPK estimates were substantially higher: 2.55 L/kg (equivalent to 0.714 L or 714 mL) for muscle and 0.856 L/kg (0.240 L or 240 mL) for liver. Similarly, physiological plasma flows are reported to be around 925 mL/h for muscle and 21.1 mL/h for liver, whereas our estimated values were Q _in,muscle = 2.09 L/h/kg (≈585 mL/h), Q _out,_muscle = 14.3 L/h/kg (≈4004 mL/h), Q _in,liver = 2.5 L/h/kg (≈700 mL/h), and Q _{out,liver,healthy} = 218 L/h/kg (≈61,040 mL/h). These differences indicate that the PopPK parameters should not be interpreted as direct representations of physiological volumes or flows, but rather as empirical descriptors optimized to capture the observed pharmacokinetic behavior.

Variability was included in CL, Q ₁, V ₃, Q _out,liver and V ₄, and the presence of diabetes was explored as a covariate in these parameters. The intercompartmental clearances Q ₁ and Q _out,liver were influenced by diabetes, with parameters estimated 8 times greater for Q ₁ (0.338 and 2.61 L/h/kg) and 11 times smaller for Q _out,liver (218 and 19.7 L/h/kg) in animals with diabetes in comparison to healthy. Besides total plasma and free muscle exposure did not differ between healthy and diabetic rats; free liver concentrations of VDG in healthy animals were lower than those observed in rats with induced diabetes. The diabetes influence on drug distribution to tissues can be attributed to the reported alteration in tissue permeability and blood flow in the diabetic state [35, 36], justifying the smaller Q _out,liver estimated for these animals. To our knowledge, this is the first popPK model reported to describe VDG total plasma and free tissue concentrations in relevant tissues in both healthy and diabetic rodents, demonstrating the impact of the disease on this antidiabetic distribution.

Using allometry and the absorption parameters of Landersdorfer et al. [24], free concentrations of VDG in muscle and liver of patients with diabetes were estimated following doses between 25 and 100 mg given once or twice daily. These free concentrations were used to estimate enzyme inhibition in both tissues where VDG exerts its effect by inhibiting DPP‐IV, the enzyme responsible for degrading GLP‐1 and GIP, prolonging the action of these incretins and reducing glucose levels. DPP‐IV inhibition after VDG administration in an oral glucose tolerance test (OGTT) in patients with T2DM was modeled with a PD I _max model, with an IC₅₀ of 1350 mg/L [10]. The same PK/PD model was used to estimate the enzyme inhibition using the estimated free tissue concentrations.

Vildagliptin efficacy can be evaluated by the level of DPP‐IV inhibition. He et al. [10] evaluated VDG total plasma concentrations and the effect of this enzyme inhibition on the plasma levels of glucose, insulin, GLP‐1, GIP, and glucagon levels, also in plasma. The authors stated that the enzyme inhibition must exceed at least 70% to achieve clinically relevant effects [28]. A study performed in mice correlated the maximum efficacy of DPP‐IV inhibitors increasing from 2 to 3‐fold GLP‐1 levels when the enzyme is 84% inhibited during the entire dosing interval [37, 38], establishing a minimum of 80% of DPP‐IV inhibition during meal intake for the onset of the effect. Moreover, an association between DPP‐IV inhibition and the efficacy of DPP‐IV inhibitors, based on a reduction in HbA1 levels, indicates the need to have an average enzyme inhibition of 91.6% over the dosing interval to achieve clinical efficacy [27]. Taking into consideration the most rigorous criteria, we choose 80% and 92% enzyme inhibition during the entire dosing interval as cut‐offs to evaluate the VDG efficacy in the present study.

The concentration‐time simulations plotted in Figure 5A,B demonstrated that 25, 50, and 100 mg resulted in plasma concentrations above the IC₅₀ when the doses were administered once or twice a day. However, in muscle and liver, the IC₅₀ was only achieved for 25, 50, and 100 mg when doses were administered twice daily, or for 100 mg once daily. The IC₅₀ indicates only half of the maximum enzyme inhibition and the treatment efficacy is dependent on > 80% inhibition. The PK/PD simulations relating VDG free concentrations and DPP‐IV inhibition over time according to the I _max model (Figure 5C,D) indicate that although plasma concentrations could indicate efficacious treatment for doses between 25–100 mg q12 h and 100 mg once a day, treatment tissue efficacy is only reached by shortening the dosing interval for the higher dose. As seen in Table 2, only 100 mg administered twice daily ensures a higher percentage of time of > 80% enzyme inhibition in muscle. None of the dosing regimens generated efficacious free VDG liver concentration. Furthermore, 92% enzyme inhibition for the entire duration of the dosing interval was only achieved by free plasma concentrations following 50 and 100 mg q12 h. In tissues, none of the dosing regimens achieved this efficacy criterion. A clinical trial that evaluated the noninferiority between vildagliptin and glimepiride treatment in 2‐years follow‐up showed only 36.9% of patients achieved the desired reduction in Hba1 compared to baseline, which was < 7% if baseline was > 7%, or less than 6.5% if baseline was greater than 6.5% [39]. The clinical endpoint suggests that not every patient undergoing vildagliptin treatment for T2DM achieves the desired HbA1c reduction. These findings corroborate with our proposal to evaluate VDG tissue levels in diabetic patients to better select efficacious treatment to treat T2DM.

Some assumptions were made to generate results useful for clinical practice. We assumed that diffusion was the process that governed VDG drug distribution. The literature reports that VDG is a substrate to the efflux transport P‐glycoprotein (P‐gp) and that this transporter plays a role in VDG absorption [25, 40], with a Michaelis–Menten affinity constant of 500 μmol/L (approximately 151.7 mg/L) [25]. In the present study, VDG was administered intravenously to mice and saturable transport from plasma to tissues was not necessary to describe the data, probably because the measured concentrations were 15 times lower than the VDG affinity constant to P‐gp. For clinical simulations was assumed that pharmacokinetic parameters remained constant over the simulation period. Changes in tissue concentration levels simulated for the disease scenario were incorporated into the pharmacodynamic model, and a PD model of DPP‐4 inhibition was similarly developed for this population. Regarding the PK–PD relationship, potential feedback from the PD component on VDG disposition was not included in the model, as modulation of DPP‐4 activity resulting from VDG inhibition is anticipated to have minimal impact on insulin secretion by pancreatic β‐cells in the absence of postprandial elevations in plasma glucose originating from intestinal absorption [41]. However, it is important to highlight that for vildagliptin, target binding plays a significant role in its pharmacokinetic behavior, particularly in tissue distribution. Vildagliptin forms a reversible covalent bond with the DPP‐4 enzyme, which is widely expressed on cell membranes and also present in soluble form in plasma. This interaction can lead to prolonged tissue retention in DPP‐4–rich organs such as the liver, kidney, and intestine, influencing its apparent volume of distribution and elimination kinetics. Thus, the extent and reversibility of target binding contribute to both the pharmacokinetic profile and pharmacodynamic duration of vildagliptin, illustrating a mild target‐mediated drug disposition (TMDD) effect [42]. In the model, intercompartmental clearance (Q _1,healthy and Q _1,diabetic) showed marked differences between healthy rats (0.338 L/h/kg) and diabetic rats (2.61 L/h/kg), suggesting that pathophysiological changes may alter redistribution dynamics. Notably, intercompartmental clearance out from the liver (Q _out) was extremely high in healthy animals (Q _{out,liver,healthy} 218 L/h/kg) but decreased significantly in diabetic animals (Q _{out,liver,diabetic} 19.7 L/h/kg), indicating greater tissue retention under disease conditions.

Another limitation of the study regards the animal model of diabetes. Chemical inducing diabetes in preclinical models are not the only way of inducing diabetes in animals, but is widely used by many researches due to the feasibility, low cost, and ability to mimic the process behind beta‐cell failure and the development of insulin deficiency [43]. For that, it was assumed that the disease effect in intercompartmental clearances was the same for rats and humans.

Nevertheless, the findings of this work demonstrated that VDG free plasma concentrations are not a good surrogate for the drug's free muscle and liver concentrations. As a result, using plasma concentrations to predict the antidiabetic effect will overestimate the actual effect expected in those tissues. Furthermore, the study showed that T2DM increased VDG free liver concentrations in comparison to the levels observed in healthy animals, reinforcing the idea of evaluating dose regimens in disease conditions, where in this case, the effect on the liver can be higher than expected.

As it was shown, the different therapeutic regimens impact drug effect, with better outcomes predicted for the administration twice daily. Based on the two criteria used to evaluate VDG effect, none of the simulated dosing regimens achieved the required level of inhibition in both tissues (liver and muscle) throughout the entire dosing interval. These findings need to be validated in the clinic, in addition to the need for further research to optimize VDG treatment strategies for treating T2DM.

Author Contributions

B.V.A. designed the research; B.B.D. and L.B.O. performed the research; B.B.D. and B.V.A. analyzed the data, and B.B.D., L.B.O., and B.V.A. wrote the manuscript.

Funding

The authors acknowledge the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brazil (CAPES) for the scholarship—Finance Code 001. B.V.A. acknowledges CNPq for the productivity fellowship (315521/2023‐6).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Groups of experiments and animals used for pharmacokinetic analysis.

Table S2: PK parameters for diabetic patients estimated by allometry.

Dias B. B., Olivo L. B., and de Araújo B. V., “Preclinical Modeling and Simulation to Explore the Tissue/Plasma Exposure and Pharmacodynamic Effect of Vildagliptin in Diabetes Treatment,” CPT: Pharmacometrics & Systems Pharmacology 15, no. 2 (2026): e70165, 10.1002/psp4.70165.