CN115440316A - Hypoglycemic drug group pharmacokinetic model construction method and application thereof - Google Patents

Abstract

The invention provides a method for constructing a pharmacokinetic model of a hypoglycemic drug group, which comprises the following steps: firstly, developing a PK model to obtain individual PK parameter values of a subject, wherein the individual PK parameter values can be used for constructing a linear or nonlinear PK-PD model; the PD model is an indirect effect model driven by AUC or blood concentration. Based on domestic and foreign clinical data, pharmacokinetic characteristics of the hypoglycemic drug are explored through model construction and fitting, sources and degrees of PK variation affecting the drug in a subject population are identified and quantitatively researched, and a PD model which possibly needs to be developed is evaluated. The source and extent of PD variation affecting the drug in healthy subjects and T2DM patient populations was qualitatively and quantitatively explored.

Description

Hypoglycemic drug group pharmacokinetic model construction method and application thereof

Technical Field

The invention belongs to the field of biological medicine, and particularly relates to a method for constructing a pharmacokinetic model of a hypoglycemic drug group and application thereof.

Background

Medicine substituting kineticThe Pharmacokinetics (PK) and Pharmacodynamics (PD) respectively study the treatment process of a drug by a body and the action of the drug on the body, the PK and the PD are inseparable unified whole which are synchronously carried out in vivo, and a pharmacokinetics/pharmacodynamics (PK/PD) model combines the pharmacokinetics with the pharmacodynamics, so that people can deeply understand the in vivo process and the action process of the drug. At present, PK/PD has been extensively considered and applied in various stages of new drug development, therapeutic drug monitoring, clinical personalized medicine and the like ^[1] 。

Hypoglycemic drugs generally refer to drugs having hypoglycemic effects for diabetic patients, and the types and the number of hypoglycemic drugs on the market are numerous at present, and a plurality of novel hypoglycemic drugs and preparations thereof are also under research and development. Pharmacokinetic studies on novel drugs are an important index for evaluating drugs.

The PK/PD model integrates the information about the interaction of the medicine and the organism, so that the whole process of the medicine effect is understood from a higher level and is quantitatively described. NN2211, liraglutide, is a GLP-1 analog with a long half-life and can be administered subcutaneously 1 time per day. In phase I clinical trials of this drug, researchers have evaluated the effect of NN2211 on insulin and blood glucose in healthy humans using the PK/PD model method. The subjects were divided into 9 groups, with group 1 being a placebo control and the other 8 groups being low-to-high dose groups. Since N N2211 has no insulin secretion promoting effect on healthy people with normal blood sugar, 3m i N of intravenous infusion glucose is given 9h (T max of liraglutide) after administration to raise blood sugar, so as to observe the effects of the medicine on insulin and blood sugar. A PK/PD model is established based on the drug action and the blood sugar regulation mechanism of the body. ^[1]

Mager et al slightly modified on the basis of the P K/P D model of N N2211 to establish a PK/PD model of exenatide in healthy and diabetic people. Healthy and diabetic subjects underwent a 5h high glucose clamp test and were monitored for blood glucose and insulin concentrations throughout the course of 1-2h by intravenous drip of exenatide injection. ^[2]

The way of modelling may be similar based on different drugs, different experimental data, but the results of modelling are certainly different, each model in fact having its specificity. Particularly for different drugs, there are great differences, and few studies on the pharmacokinetic models of PB-119 and PB-201 in the prior art are available.

[1] The application of the Lixingang, zhoutian, luwein, etc. pharmacokinetic/pharmacodynamic model in the evaluation of hypoglycemic medicine [ J ] medicine evaluation 2014 (10): 6.

[2]Gabrielsson J,Jusko WJ,Alari L.Modeling of dose-response-time data:four examples of estimating the turnover parameters and generating kinetic functions from response profiles[J]. Biopharm Drug Dispos,2000,21(2):41-52.

Disclosure of Invention

In order to solve the problems, the PK behavior of the hypoglycemic drug in different races is simulated by a group pharmacokinetic method, the pharmacokinetic characteristics of the hypoglycemic drug in healthy subjects and type 2 diabetes patients (T2 DM) are explored through model fitting based on domestic and foreign clinical data, and the source and the degree of PK variation of the drug in a subject group are identified and quantitatively researched. By the validation of quantitative pharmacological models, the pharmacodynamic characteristics of hypoglycemic drugs in healthy subjects and T2DM were understood, and PD models that may need to be developed were evaluated. Qualitative and quantitative exploration was conducted to influence the origin and extent of PD variation of drugs in healthy subjects and T2DM patient populations.

On one hand, the invention provides a method for constructing a pharmacokinetic model of a hypoglycemic drug group.

The method comprises the following steps:

firstly, developing a PK model to obtain individual PK parameter values of a subject for constructing a PD model; the PD model is an indirect effect model driven by AUC and is used for describing HbA1c.

Also included is evaluating the covariate factor.

The PK model is firstly constructed into a basic model, and a final model is obtained according to the extremely significant covariates.

The PK model basic model is as follows:

where Cp refers to the drug concentration in the central compartment.

The basic model of the PD model is as follows:

HbA1c _placebo ＝HbA1c _baseline ·(e ^-HLAM·t )+HSLP·t/168

K _in ＝K _out ·HbA1c _baseline

HbA1c _drug (0)＝0

HbA1c＝HbA1c _drug +HbA1c _placebo

wherein HbA1c _placebo Refers to HbA1c values after taking placebo for patients with Mediterranean T2 DM.

And the basic model of the PD model is the final model.

The hypoglycemic agent is GLP-1 receptor agonist or glucokinase activator.

Preferably, the GLP-1 receptor agonist is PB-119.

Preferably, the glucokinase activator is PB201.

In another aspect, the invention provides a pharmacokinetic model of a population of hypoglycemic agents.

The hypoglycemic agent is PB-119; the model is a PK-PD model; the PK model is a 1-compartment model of linear elimination and absorption; the PD model is an indirect effect model driven by AUC and is used for describing HbA1c.

The basic model of the PK model is as follows:

where Cp refers to the drug concentration in the central compartment.

The basic model of the PD model is as follows:

HbA1c _placebo ＝HbA1c _baseline ·(e ^-HLAM·t )+HSLP·t/168

K _in ＝K _out ·HbA1c _baseline

HbA1c _drug (0)＝0

HbA1c＝HbA1c _arug +HbA1c _placebo

wherein HbA1c _placebo Refers to HbA1c values after placebo administration in patients with T2DM in the middle of the body.

The model also included the PK model very significant covariates, body weight and disease progression.

The quantitative relationship is as follows:

a healthy person:

patients on initial treatment (monotherapy):

patients with poor metformin control (combined metformin therapy):

a healthy person: ka = theta _Ka ·e ^ηKa

Patients on initial treatment (monotherapy): ka = theta _Ka ·0.889·e ^ηKa

Patients with poor metformin control (combined metformin therapy): ka = theta _Ka ·1.24·e ^ηKa

A healthy person: vc = θ _Vc ·e ^ηVc

Patients on initial treatment (monotherapy): vc = θ _Vc ·1.65·e ^ηVc

Patients with poor metformin control (combined metformin therapy): vc = θ _Vc ·2.54·e ^ηVc

Wherein, CL and K _a 、V _c Respectively indicating the system clearance rate, the absorption rate constant and the distribution volume;

θ _CL θ _Ka θ _Vc respectively refer to CL and K _a 、V _c A population-typical value of (a);

WT is body weight, kg;

e ^ηCL e ^ηKa e ^ηVc respectively refer to CL and K _a 、V _c The random effect of (a).

The PD model has no covariates, and the basic model is the final model.

In yet another aspect, the invention provides a pharmacokinetic model of a population of hypoglycemic agents.

The hypoglycemic agent is PB-201;

the model is a PK-PD model;

the PK model is a 1-compartment model with nonlinear absorption and linear elimination;

the PD model is an indirect effect model driven linearly by placebo effect and the central compartment PB-201 plasma drug concentration.

The basic model of the PK model is

Wherein, V _m Refers to the maximum absorption rate;

K _m refers to the amount of drug that corresponds to half the maximum absorption rate;

A _a refers to the amount of drug in the absorption compartment;

C _p refers to the concentration of drug in the central compartment;

V _c refers to the volume of distribution of the drug in the central compartment;

CL refers to the systemic clearance of the drug; h refers to relative bioavailability;

IMAX refers to the maximum percentage of reduction in bioavailability;

IC50 refers to the dose at which half the maximum percent reduction in bioavailability is achieved;

DOSE refers to the administered DOSE;

the basic model of the PK model is used for describing the blood concentration-time curve of healthy people and T2DM patients after oral administration of PB-201.

The basic model of the PD model is as follows:

wherein R is _placebo (0)＝1；Kinp _lacebo ＝Kout _placebo *R _placebo (0)

Wherein R is _drug (0)＝1；Kin _drug ＝Kout _drug *R _drug (0)

ΔFPG _drug ＝FPG _baseline *Fraction*(1-R _drug ) (7)

ΔFPG _placebo ＝FPG _baseline *(1-Fraction)*(1-R _placebo ) (8)

FPG＝FPG _baseline -ΔFPG _drug -ΔFPG _placebo (9)

HbA1c(0)＝HbA1c _baseline (11)

Wherein Placebo refers to a Placebo effect;

kp refers to the rate constant for elimination of placebo effect;

R _placebo refers to the percentage of non-phosphorylated glucose in pancreatic beta-cells and liver driven by placebo effect relative to baseline;

Kin _placebo is denoted by R _placebo The generation rate constant of (2);

Kout _placebo is denoted by R _placebo The elimination rate constant of (2);

R _drug refers to the percentage of unphosphorylated glucose in pancreatic β -cells and liver driven by drug effects relative to baseline;

Kin _drug is denoted by R _drug The generation rate constant of (2);

Kout _drug is denoted by R _drug The elimination rate constant of (2);

k means a decrease in drug concentration R _drug The effect of (c).

Drawings

FIG. 1 is a PB-119PK basic model diagnostic diagram.

FIG. 2 is a diagnostic graph of the PB-119PK final model.

FIGS. 3-6 show the final model simulation results for PB-119 PK.

FIG. 7 is a diagnostic chart of the final model of PB-119 PD.

FIG. 8 shows the final model simulation results of PB-119 PD.

Figure 9 is a graph of drug concentration versus time for 24 weeks of 5 doses of PB-119 μ g qw, 50 μ g qw, 100 μ g qw, 150 μ g qw, 200 μ g qw for the treatment naive (monotherapy) and for the treatment poorly with metformin (combination metformin therapy) patients on a continuous subcutaneous injection.

FIG. 10 is a graph of the change in HbA1c relative to baseline and placebo for 24 weeks following QW subcutaneous injections of 100 μ g, 150 μ g, 200 μ g doses in patients (upper) simulating Chinese treatment (monotherapy) and patients with poor metformin control (combined metformin therapy) (lower).

FIG. 11 is a PB-201 population PK simulation.

FIG. 12 is a PB-201 population PD simulation view.

It should be noted that, the part of the above pictures is an analysis software straight-through picture, and the individual details are not clear and difficult to modify, but do not affect the determination of the final technical effect.

Detailed Description

The present invention will be described in further detail with reference to specific examples, which are not intended to limit the present invention, but to illustrate the present invention. The experimental methods used in the following examples are not specifically described, and the materials, reagents and the like used in the following examples can be obtained commercially in general under the conventional conditions without specific descriptions.

Example 1 method for constructing pharmacokinetic model of hypoglycemic drug population and application thereof

The hypoglycemic agent in the embodiment is PB119, PB-119 taken by all subjects is injection (1 mL/0.5 mg), each batch meets the product quality standard, and the specific specification, batch number and clinical application condition are shown in the following table:

the invention provides a clinical pharmacology basis for determining a clinical study treatment window and recommending an optimal treatment dose for phase III clinical study by analyzing the completed pharmacokinetic/pharmacodynamic (PK/PD) characteristics of Chinese phase I (ICP-I-2013-08, ICP-I-2014-07 and ICP-I-2015-01) and American phase I study (CSP-PB 119-US 01-01) and Chinese phase II (PB 119201) and American phase II (PB 119202) and related demographic and pathophysiological factors by using a population method, exploring PK/PD characteristics of PB-119 in Chinese healthy subjects and patients with type 2 diabetes mellitus (T2 DM), identifying and quantitatively researching the source and degree of PK/PD variation, and combining an exposure/safety relation. The 5 subjects studied were all Chinese healthy subjects of male or female not less than 18 years of age, or patients with T2DM in both China and China, and the Body Mass Index (BMI) was 18.5-40kg/m ² In the meantime.

Study ICP-I-2013-08 was a single-center, randomized, double-blind, placebo-controlled, single-dose escalation study of 1 in a total of 70 healthy volunteers enrolled in 54 subjects using the test drugs in chinese healthy subjects.

Study ICP-I-2014-07 is 1 single-center, randomized, open, multiple dose escalation study in chinese healthy subjects with a total of 36 healthy volunteers in the group and 36 subjects using the test drug.

Study ICP-I-2015-01 was 1 single-center, randomized, open, positive drug control parallel control study in chinese T2DM patients, for a total of 36 male and female T2DM patients into groups, and 24 subjects used test drugs.

Study CSP-PB119-US01-01 was 1 randomized, double-blind, placebo-controlled, multiple dose escalation study in u.s.t 2DM patients, 40 patients with T2DM enrolled, and 32 subjects using the test drug.

Study PB119201 is a multicenter, randomized, double-blind, placebo, parallel control study of 1 in chinese untreated T2DM patients, 251 male and female T2DM patients were enrolled, 188 subjects were on trial medication, and 1 subject was randomized without trial medication.

Study PB119202 was a multicenter, randomized, double-blind, placebo-controlled 12-week treatment phase II study conducted in the united states with 217 male and female T2DM patients enrolled, 163 subjects using the test drugs, in patients with poorly controlled metformin monotherapy.

The analysis method comprises the following steps:

at various stages of model development, the model estimation method used was First Order Conditional Estimation (FOCEI) to evaluate the interaction. Data exploratory analysis was performed using R (v 3.5.3) or R (v 3.6.2) software, and simulations and observations were performed using NONMEM m 7.2.

The software runs on the Windows Server 2016Standard operating system.

1. Analyzing a data set

Data are expressed in "median (minimum, maximum)".

Population PK study data, age 45.0 (19.0, 70.0) years, body weight 68.7 (49.0, 130.8) kg, BMI 25.1 (19.5, 39.8) kg/m ² The sex ratio of men and women is about 1.00:1.04 and a Body Surface Area (BSA) of 1.76 (1.43, 2.50) m ² The creatinine clearance was 77.8 (26.5, 113) μmol/L.

Population PD study data, T2DM patients were 56.0 (26.0, 70.0) years of age, 82.6 (56.6, 130.8) kg body weight, and 29.2 (23.7, 39.8) kg/m BMI ² Male and female sex is about 1.30:1.00 Fasting Plasma Glucose (FPG) at baseline of 9.62 (5.10, 15.7) mmol/L, and glycosylated hemoglobin (HbA 1 c) at baseline of 8.25 (7.00, 10.0)%; the placebo group subjects were 55.5 (23.0, 75.0) years of age, 79.0 (47.2, 130.8) kg body weight, and 28.4 (20.3, 40.5) kg/m BMI ² Male and female gender was about 1.35: 1.00 Fasting Plasma Glucose (FPG) at baseline was 9.46 (5.10, 22.0) mmol/L and glycosylated hemoglobin (HbA 1 c) at baseline was 8.45 (6.00, 10.6)%.

2. Modeling process

The PB-119 population PK/PD model analysis was performed in the following six steps:

(1) Exploratory data analysis

Exploratory data analysis and data visualization analysis are used to understand the intrinsic information of the data, to find extrema and potential outliers, to evaluate the likely trends of the data and to determine if a data set is missing. Individual drug time curves (constant coordinates, logarithmic graphs) are plotted for PK/PD data, trough concentration time curves, PD indices (FPG, hbA1 c) are plotted heuristically over time, demographic information distributions are plotted, and extrema and potential outliers of the relevant data are determined. And determining continuous covariates related to the parameters based on the covariate pairing graph, judging whether the covariates have collinearity, determining classified covariates related to the parameters according to the box line graph, and providing basis for screening the subsequent covariates.

And (3) judging abnormal values:

outliers were detected in the complete PK or PD curves by individual and overall drug concentration or PD index-time curves. At concentration points in the curve that deviate significantly from the neighboring values or from the values of other subjects at the same time point (diagnosis shows | CWRES | > 6), it should be determined whether to mark this point as an outlier and to cull it from the bulk PK/PD analysis set, the PK data set has no data points for | CWRES | >6, and the PD data set has culled 6 data points for | CWRES | > 6. If there is an abnormal PK/PD behavior for the entire PK/PD profile for an individual subject and there is no clear reason (e.g., combination, non-compliance with treatment regimen data, post-dose discharge, etc.) to explain the data abnormality, all data for that subject should be rejected and this is not the case in this study.

Actual analysis of population PK studies 163 subjects, of which ICP-I-2013-08 studies 36; ICP-I-2014-07 studied 35, 1 subject AE regressed, and the original data had no PK information of this subject; 24 cases of ICP-I-2015-01 were studied; CSP-PB119-US01-01 study 30, 2 subjects detected exenatide in the pre-dose sample, and were considered to have been exposed to the drug prior to the trial and therefore were not included in the analysis; PB119202 study 38, 7 subjects advanced group withdrawal during the trial, with all PK data missing.

Population PD study actual analysis 178 subjects, of which ICP-I-2015-01 study 24; PB119202 study 38 trial drugs and 54 placebo, early withdrawal during 7 subject trials, all PK data missing and not included in the PD analysis dataset; PB119201 study 62 placebo.

(2) Base model development

A sequential analysis method (sequential Modeling) is adopted between PK and PD models for Modeling research. The method is characterized in that a PK model is firstly developed, individual PK parameter values of a subject are obtained, and the PK parameter values are used for constructing a PD model, so that clinical pharmacological support is provided for the optimal design of a dosing scheme.

Model analysis was performed using NONMEM7.2, analysis was performed using PIRANA 2.8.0 calling the NONMEM7.2 program, and preliminary mapping was performed calling R/XPOSE.

This example investigated a total of 3546 concentration points of PB-119 for population PK modeling analysis, with a baseline model diagnostic profile as shown in figure 1. The results of the population PK model analysis show that a linear absorption, linear elimination 1-compartment model best describes the PK profile of PB-119 in chinese healthy subjects, chinese T2DM patients and us T2DM patients over a dose range of 25-500 μ g.

The PK basal model is: the plasma concentration-time curve of healthy people and T2DM patients after subcutaneous injection of PB-119 can be better described by using a 1-compartment model of linear elimination and absorption, and a PK basic model is as follows:

where Cp refers to the drug concentration in the central compartment.

The PD basic model is: an indirect effect model (inhibition of HbA1c production) driven by AUC quantifiably describes the characteristics of HbA1c after 25-500 μ g dosing in middle american T2DM patients, including chinese untreated T2DM and u.s.a. Metformin poorly controlled T2DM patients; the exponential equation with placebo effect and the linear equation of the effect of disease progression can better describe the pharmacodynamic properties of HbA1c in Chinese T2DM patients and in American T2DM patients after placebo, and the basic model equation describing HbA1c is as follows:

HbA1c _placebo ＝HbA1c _baseline ·(e ^-HLAM·t )+HSLP·t/168

K _in ＝K _out ·HbA1c _baseline

HbA1c _drug (0)＝0

HbA1c＝HbA1c _arug +HbA1c _placebo

wherein HbA1c _placebo Refers to HbA1c values after placebo administration in patients with T2DM in the middle of the body.

(3) Modifying a base model

(4) Evaluating covariate factors

Based on a basic model, a correlation coefficient matrix is obtained through correlation analysis calculation, a correlation coefficient graph is drawn, variables with correlation coefficients larger than or equal to 0.3 are screened out to be used as potential covariates to be investigated, correlation analysis results among the potential covariates show that the correlation coefficients among the covariates are smaller than 0.8, and no collinearity exists among the covariates.

To prevent the effects of co-linearity and confounding factors of the covariate model, the correlation between covariates is considered. There are highly correlated covariates (e.g., body weight and lean body mass), and only 1 covariate can be used for model evaluation based on a one-factor analysis of the likelihood and degree of correlation of the correlation mechanism of the two parameters. Two or more highly correlated covariates cannot be included in the covariate model for the same parameter. The present study did not incorporate two highly correlated covariates on one parameter at the same time.

Covariates were added using the forward method (< 0.05, MVOF change > 3.84 with df = 1) and removed using the backward method (< 0.001, MVOF change < 10.83 with df = 1).

Pharmacokinetic covariate factors evaluated included: age, height, weight, BMI, body surface area, race, sex, glutamic-pyruvic transaminase, glutamic-oxaloacetic transaminase, albumin, glutamyl transpeptidase, blood creatinine, triglycerides, total cholesterol, lean body mass, ideal body weight, creatinine clearance (CG equation calculation and MDRD equation calculation), anti-drug antibodies, demographic characteristics (i.e., disease progression, divided into healthy people, treatment naive patients (monotherapy) and patients with poor metformin control (combined metformin therapy)).

Pharmacodynamic covariate factors evaluated included: demographics (weight, age, sex, BMI), race, triglyceride, total cholesterol, fasting insulin baseline, C-peptide baseline, insulin sensitivity (calculated by the HOMA-IR equation), beta cell secretory capacity (calculated by the HOMA-B equation).

PK model covariates:

after the analysis of the covariates of the plan analysis in a step-by-step manner, the results show that the body weight has a very significant influence on the Clearance (CL) and the disease progression respectively on the Clearance (CL), the distribution volume (Vc) and the absorption rate constant (KA) (p is less than 0.001).

When the other covariates were unchanged, the steady state AUC increased by 50.1% or decreased by 53.4%, the steady state trough concentration increased by 60.3% or decreased by 56.5%, and the steady state peak concentration increased by 36.3% or decreased by 47.5% when the body weight changed from the median of 68.7kg to the extremum of this study, 49.0 or 130.8 kg; when the disease state changed from a healthy person (DIS = 0) to a patient treated initially (monotherapy) (DIS = 1) or with poorly controlled metformin (combined metformin therapy) (DIS = 2), the peak concentration at steady state decreased by 27.6% or 42.6%, the AUC at steady state decreased by 15.5% or 45.0%, and the trough concentration at steady state decreased by 4.45% or 51.2%. The above results indicate that body weight or disease state (combined metformin therapy) has some effect on PB-119 exposure in patients with middle american T2 DM.

Finally 2 statistically significant covariates were identified for weight and disease progression with the following quantitative relationships:

a healthy person:

patients on first treatment (monotherapy):

patients with poor metformin control (combined metformin therapy):

a healthy person: ka = theta _Ka ·e ^ηKa

Patients on initial treatment (monotherapy): ka = theta _Ka ·0.889·e ^ηKa

Patients with poor metformin control (combined metformin therapy): ka = theta _Ka ·1.24·e ^ηKa

A healthy person: vc = θ _Vc ·e ^ηVc

Patients on first treatment (monotherapy): vc = θ _Vc ·1.65·e ^ηVc

Patients with poor metformin control (combined metformin therapy): vc = θ _Vc ·2.54·e ^ηVc

Wherein, CL and K _a 、V _c Respectively indicating the system clearance rate, the absorption rate constant and the distribution volume;

θ _CL θ _Ka θ _Vc respectively refer to CL and K _a 、V _c A population typical value of (a);

WT is body weight, kg;

e ^ηCL e ^ηKa e ^ηVc respectively refer to CL and K _a 、V _c The random effect of (a).

PD model covariates:

covariates that may affect PD parameters were introduced and culled, and eventually no covariates with statistically significant significance were found.

(5) Final full-scale model correction

For the full-scale model, which only retains very significant covariates, the inter-individual variation and residual error models need to be evaluated. The covariate equations of the model should be as simple as possible, e.g., the power function term parameters reduce to 1 or significantly eliminate the covariate group, which can be redefined with fewer groups or parameters. The optimization fit diagnostic map is also used to test the model.

(6) Final model evaluation

Final PK model:

they were evaluated using a simulation-based visual predictive test method (VPC) and Bootstrap method (boottrap). Performing 1000 times of simulation on the final model by using NONMEM7.2 software, and comparing a result calculated according to 1000 times of simulation data with an actually measured result; a 5th,50th (median) 95th percentile distribution-time plot of the simulated-based predicted concentrations is plotted, while the measured data and/or the percentile distribution based on the measured data is plotted overlappingly (i.e., a VPC plot) to evaluate the agreement between the model-based data and the measured data. The original dataset was sampled 500 times and the results of the 90% Confidence Intervals (CI) of the parameters calculated from the 500 bootstrap datasets were compared to the observed results.

Final PK model parameters are shown in the table below.

NA: not applicable;

IIV: inter-individual variation;

CL: central compartment clearance rate;

vc: a distribution volume of the central chamber;

KA: an absorption rate constant;

CL-WT: the effect of body weight as a covariate on CL;

CL-DIS-1: treatment naive (monotherapy patients as a covariate effect on CL;

CL-DIS-2: the effect of patients with poor metformin control (combined metformin therapy) on CL as a covariate;

VC-DIS-1: the effect of naive (monotherapy) patients as covariates on Vc;

VC-DIS-2: the effect of patients with poor metformin control (combined metformin therapy) on Vc as a covariate;

Ka-DIS-1: the effect of treatment naive (monotherapy) patients as covariates on Ka;

Ka-DIS-2: effect of patients with poor metformin control (metformin treatment combined) on Ka as a covariate;

epsilon: and (6) model residual errors.

Compared with the basic model, the final model can improve the model goodness of fit, and compared with the basic model, the PK parameter IIV is reduced after the covariate is added, wherein the IIV of CL/F is reduced by about 25 percent (82.6 percent is reduced to 57.2 percent), and the IIV of VC is reduced from 94.2 percent to 85.0 percent, which shows that the addition of the covariate can partially explain the inter-individual variation of the parameters. Final model diagnosis fig. 2 shows that the final PopPK model can more accurately describe the pharmacokinetic profile of PB-119 in healthy persons in the middle and beauty and patients.

Final PD model:

the study finally included a total of 763 HbA1c concentration points for PD modeling.

Covariate analysis of the PD fundamental model does not identify covariates which have significant influence on PD, the fundamental model is the final model, and the final population PD model parameter estimation value (Run 008) is shown in the following table:

NA: not applicable;

K _out : rate constant of elimination of HbA1c in middle american T2DM patients;

SLOPE1 and SLOPE2: the inhibition constant of PB-119 to HbA1c in Chinese T2DM and American T2DM patients;

HBL: a baseline value of glycated hemoglobin (HbA 1 c);

HLAM1 and HLAM2: rate constants for placebo effect of HbA1c after treatment for a certain time in Chinese T2DM patients and in US T2DM patients, respectively;

HSLP1 and HSLP2: respectively refers to HbA1c disease process effect of Chinese T2DM patients and American T2DM patients after receiving treatment for a certain time;

epsilon: proportional residual of HbA1c.

Finally, the PD model searched only for the HbA1c change model, and no body weight model was searched (the differences in clinical data in china and america were large).

3. Population PK/PD model simulation

PK simulation protocol:

based on the final PK model, 468 patients with combined middle american phase II population characteristics (china n =251, us n = 217) were simulated as 25 μ g, 50 μ g, 100 μ g, 150 μ g, 200 μ g dose QW subcutaneous PB-119 drug injection curves for naive patients (monotherapy) and patients with poor metformin control (combined metformin therapy), respectively. Sensitivity analysis was further performed on covariate body weight and disease status (healthy, naive and patients with poor metformin control) that significantly affected PK. The pharmacokinetic profile of T2DM patients (including Cmax, ss, cmin, ss, aucs) with body weights varying from median 68.7kg to extreme 49.0kg or 130.8kg in this study was simulated while other covariates were unchanged to quantitatively describe the extent to which significant covariates affected the pharmacokinetic profile. The pharmacokinetic profile (including Cmax, ss, cmin, ss, aucs) of T2DM patients with healthy humans (DIS = 0) to single drug therapy (DIS = 1) or combined metformin therapy (DIS = 2) was also simulated to quantitatively describe the extent to which significant covariates affect the pharmacokinetic profile.

The final model was simulated 1000 times using the NONMEM7.2 software, and the results are shown in FIGS. 3-6, and it can be seen from FIGS. 3-6 that the structure of the PopPK final model has no obvious deviation, and the pharmacokinetic characteristics of PB-119 in the healthy people and patients in the middle and the beauty can be estimated more accurately. Compared with the typical subject, namely a healthy subject with the weight of 67.8kg, a healthy subject with the weight of 49.0kg or 130.8kg, the AUC is increased by 50.1% or is reduced by 53.4%, the peak concentration at steady state is increased by 36.3% or is reduced by 47.5%, and the trough concentration at steady state is increased by 60.3% or is reduced by 56.5%; when the body weight is unchanged, patients treated initially (monotherapy) or patients with poor metformin control (combined metformin therapy) have a 15.5% or 45.0% decrease in AUC, a 27.6% or 42.6% decrease in peak concentration at steady state, and a 4.45% or 51.2% decrease in trough concentration at steady state, as compared to healthy subjects, but the effect is evaluated for clinically significant efficacy and safety changes based on the E-R relationship between efficacy and safety to determine whether adjustments to the dosing regimen are needed for patients of different body weights or disease states.

Internal validation by Bootstrap method, minimized success 58.6%, median of 500 Bootstrap results similar to parameter estimates, and 90% CI contains parameter estimates, indicating that the established model is robust. Bootstrap results are detailed in the table below.

Boostrap model Boostrap500 results population PK (boost _ dir 4)

NA: not applicable;

IIV: inter-individual variation;

CL: central compartment clearance rate;

vc: respectively, the distribution volume of the central chamber;

KA: an absorption rate constant;

CL _D -LBW: liposome weight as covariate pair CL _D The influence of (a);

CL-DIS-1: the effects on CL in treatment naive (monotherapy) diabetics;

CL-DIS-2: the effect of CL in diabetic patients with poor metformin control (combined metformin therapy);

V _C -DIS-1: the effect of diabetic patients on Vc in the initial treatment (monotherapy);

V _C -DIS-2: the effect on Vc in diabetic patients with poor metformin control (combined metformin therapy);

KA-DIS-1: the effect on KA of treatment naive (monotherapy) diabetics;

KA-DIS-2: the effect on KA of diabetic patients with poor metformin control (combined metformin therapy);

epsilon: and (6) model residual errors.

PD simulation mode:

based on the final PD model, population characteristics of all patients (china n =251, american n = 217) pooled from the middle american phase II clinical study, both on first treatment (monotherapy) and on metformin poorly controlled (combined metformin therapy) were simulated to 100 μ g, 150 μ g, 200 μ g dose QW administered subcutaneously for 12 weeks and 24 weeks, respectively, using the AUC means of each dose group of middle american patients to estimate the change in HbA1c concentration at different doses. The results are shown in the 200 μ g dose range, and PB-119 shows a dose-dependent decrease in HbA1c (FIG. 10). After Chinese treatment patients (single-drug treatment) continuously injected with PB-119 subcutaneously for 12 weeks, hbA1c mean values of HbA1c of each dose group are reduced by 0.425%,0.638% and 0.850% respectively compared with baseline and placebo; after the Chinese primary treatment patient (single-drug treatment) continuously injects PB-119 subcutaneously for 24 weeks, the HbA1c mean value of each dose group is reduced by 0.506%,0.759% and 1.01% compared with the baseline period and the placebo respectively; in patients with poor metformin control in the United states (combined metformin therapy) HbA1c after continuous subcutaneous PB-119 weeks was reduced by 0.336%,0.461%,0.716% of the mean HbA1c in each dose group compared to baseline and placebo. In patients with poor American metformin control (combined metformin therapy) after continuous subcutaneous injections of PB-119 for 24 weeks, the mean HbA1c values of the respective dose groups were reduced by 0.400%,0.549% and 0.853% compared with baseline and placebo, respectively.

The final model diagnosis is shown in fig. 7, which shows that the final PopPD model can more accurately describe the pharmacodynamic characteristics of PB-119 in the middle-american patient.

Internal validation by the Bootstrap method shows that the minimization success rate is 96.2%, the median of 500 Bootstrap results is similar to the parameter estimation value, and 90% CI comprises the parameter estimation value, which indicates that the established model is stable. Bootstrap results are detailed in the table below.

Population PD Total model Boostrap500 times results (boost _ dir 1)

NA: not applicable;

kout: the elimination rate constant of HbA1c in middle and american T2DM patients;

SLOPE1 and SLOPE2: the inhibition constant of PB-119 to HbA1c in Chinese T2DM and American T2DM patients;

HBL: a baseline value of glycated hemoglobin (HbA 1 c);

HLAM1 and HLAM2: the rate constants of the placebo effect of HbA1c after a certain period of treatment in Chinese T2DM patients and in US T2DM patients, respectively;

HSLP1 and HSLP2: the HbA1c disease course effect is respectively shown after Chinese T2DM patients and American T2DM patients receive treatment for a certain time;

epsilon: proportional residual of HbA1c.

The VPC evaluation of the final model was performed and 1000 simulations of the final model were performed using the NONMEM7.2 software, the results of which are detailed in fig. 8, and it can be seen from fig. 8 that the PD final model structure was not significantly biased and the pharmacodynamic characteristics of PB-119 in the middle and american patients could be estimated unbiased.

4. Exploration of ethnic differences

PK model studies included chinese healthy subjects, chinese T2DM patients, and us T2DM patients, and the impact of race (0 = asian, 1= caucasian, 2= black) on PK exposure was analyzed by exploratory analysis and model covariates, with race factors not included in the final PopPK model. However, since analysis of ethnicity differences is influenced by confounding factors of disease states, it is not possible to clearly distinguish whether exposure differences are caused by ethnicity or disease states, and studies with similar drugs have not found that ethnicity influences pharmacokinetics, and the main in vivo elimination mechanism of this drug is renal clearance, and it is presumed that PB-119 has a low possibility of pharmacokinetic ethnicity differences. The PopPD study included untreated T2DM patients in china and T2DM patients with poor metformin control in the united states, and the influence of ethnicities (asians, caucasians, and blacks) on PD indices was analyzed by exploratory analysis and model covariates, and finally no significant influence of ethnicities on PopPD model PD parameters was found.

5. PopPK model sensitivity analysis

In the PopPK modeling process, a point of a basic model appears in CWAS >6, the comprehensive consideration influence is not large, so that the point is not removed in the modeling process, the influence of the point is verified based on the final PopPK model, the comparison result of the model result obtained after the point is removed and the previous basic model is shown in the table below, the change of all parameters is less than 5%, and the final model result is not obviously influenced.

Comparison of population PK Total model parameter estimates (Run 139 and Run 151)

NA: not applicable;

IIV: inter-individual variation;

CL: central compartment clearance rate;

vc: a distribution volume of the central chamber;

KA: an absorption rate constant;

CL-WT: the effect of body weight as a covariate on CL;

CL-DIS-1: treatment naive (monotherapy patients as a covariate effect on CL;

CL-DIS-2: effects of metformin poorly controlled (combined metformin therapy) patients on CL as a covariate

VC-DIS-1: the effect of naive (monotherapy) patients as covariates on Vc;

VC-DIS-2: effect of metformin poorly controlled (combined metformin therapy) patients on Vc as a covariate;

Ka-DIS-1: effect of treatment naive (monotherapy) patients as covariate on Ka;

Ka-DIS-2: effect of patients with poor metformin control (combined metformin therapy) on Ka as a covariate;

epsilon: and (6) model residual errors.

6. Conclusion

The study was modeled and simulated based on a 1-compartment model of linear absorption, linear elimination, and an indirect effect model (taking placebo effect into account) of AUC driven HbA1c (inhibition of HbA1c production) changes over the 25-500 μ g dose range, leading to the following conclusions:

1) Ethnicity may have no significant impact on pharmacokinetics;

2) The pharmacokinetic differences in the middle and american clinical trials may be mainly due to disease severity (for CL/F, american poor metformin control of T2DM patients > chinese naive T2DM patients > chinese healthy);

3) In eGFR>45mL/min, under 70 years old or with a BMI of 18.5-40.0kg/m ² In T2DM patients within the range, kidney function, age or BMI had no significant effect on PB-119 exposure;

4) Body weight had a significant effect on CL/F, with body weight reduced to 49.0kg or increased to 130.8kg, cmax, ss, AUCss or Cmin based on 67.8kg body weight, with ss changes less than 60.4%;

5) Based on the exposure-HbA 1c quantification relationship, it was estimated that for 24 weeks (QW) with continuous dosing in the 100-200 μ g dose range, the HbA1c reduction increased and tended to saturate with increasing PB-119 dose in both naive T2DM patients (monotherapy) and in poorly controlled T2DM patients with metformin (combination with metformin);

6) No potential covariates were found to significantly affect PD.

Example 2 method for constructing pharmacokinetic model of hypoglycemic drug population and application thereof

The hypoglycemic agent in the embodiment is PB201. In this example, the basic steps of model construction can be referred to in example 1 unless otherwise specified.

Information about samples in phase I of the world

PB-201 has completed 1 phase I study in the home, and the results show that PB-201 50/50mg, 100/50mg and 100/100mg pre-breakfast and pre-lunch dosing changes the dynamic blood glucose index in a dose-dependent trend. In order to quantitatively describe the influence of different administration modes of PB-201 on the dynamic blood glucose index, the study uses the dynamic blood glucose as a pharmacodynamic index to carry out group pharmacokinetic and pharmacodynamic analysis.

The influence of the dynamic daily fluctuation rhythm of blood glucose, meal and administration time on blood glucose fluctuations in PB-201 treated type 2 diabetic patients (T2 DM) was quantitatively explored by analyzing the relation between pharmacokinetics and pharmacokinetics (dynamic blood glucose) of a completed phase I clinical study in China using a population method. The study aims to support the design of a PB-201 administration mode and evaluate the blood glucose fluctuation of the PB-201 in the morning and evening administration mode and the morning and afternoon administration mode.

The key assumptions for this study were: the intrinsic circadian rhythm of blood glucose levels in patients with T2DM follows the characteristic of cosine curve fluctuations, on the basis of which fluctuations in blood glucose are only influenced by the therapeutic drug PB-201 and by meals.

The modeling method and the software are different from the pharmacological model and modeling basis of 119, and the dynamic blood sugar is used as the pharmacodynamic index for the first time to perform modeling analysis.

PK model development:

in an overseas study of PB-201, a dose-dependent non-linear absorption and linear elimination 1-compartment model was used for the drug population PK model to describe plasma concentration-time profiles following oral PB-201 in healthy humans and T2DM patients, as follows:

wherein, V _m Refers to the maximum absorption rate;

K _m refers to the amount of drug that corresponds to half the maximum absorption rate;

A _a refers to the amount of drug in the absorption compartment;

C _p refers to the concentration of drug in the central compartment;

V _c refers to the volume of distribution of the drug in the central compartment;

CL refers to the systemic clearance of the drug; h refers to relative bioavailability;

IMAX refers to the maximum percentage of reduction in bioavailability;

IC50 refers to the dose at which half the maximum percent reduction in bioavailability is achieved;

DOSE refers to the DOSE administered.

Developing a PD model:

an indirect effect model (promoting FPG elimination) linearly driven by placebo effect and central compartment PB-201 plasma drug concentration best describes the FPG pharmacodynamic profile after placebo and after oral administration in subjects. Furthermore, an indirect effect model (promoting HbA1c production) driven linearly with FPG can best describe the HbA1c pharmacodynamic characteristics of a subject following oral administration. The basic model equations describing FPG and HbA1c are as follows:

wherein R is _placebo (0)＝1；Kin _placebo ＝Kout _placebo *R _placebo (0)

Wherein R is _drug (0)＝1；Kin _drug ＝Kout _drug *R _drug (0)

ΔFPG _drug ＝FPG _baseline *Fraction*(1-R _drug ) (7)

ΔFPG _placebo ＝FPG _baseline *(1-Fraction)*(1-R _placebo ) (8)

FPG＝FPG _baseline -ΔFPG _drug -ΔFPG _placebo (9)

HbA1c(0)＝HbA1c _baseline (11)

Wherein Placebo refers to a Placebo effect;

kp refers to the rate constant for elimination of placebo effect;

R _placebo refers to the percentage of non-phosphorylated glucose in pancreatic beta-cells and liver driven by placebo effect relative to baseline;

Kin _placebo is denoted by R _placebo The generation rate constant of (2);

Kout _placebo is denoted by R _p1acebo The elimination rate constant of (2);

R _drug refers to the percentage of unphosphorylated glucose in pancreatic β -cells and liver driven by drug effects relative to baseline;

Kin _drug is denoted by R _drug A generation rate constant of (a);

Kout _drug is denoted by R _drug The elimination rate constant of (2);

k means a reduction in drug concentration R _drug The effect of (c).

PK simulation protocol:

carrying out PK characteristic simulation and verification on Chinese population based on a final PK model, referring to covariate data of Chinese phase I clinical study subjects, and simulating PB-201 drug-taking time curves of 50/50mg, 50/100mg and 100/100mg doses continuously for 7 days under the condition that 23 virtual subject populations with the same population characteristics are not co-administered with ketoconazole and are administered at the morning lunch. Non-compartmental analysis (Linear Up Log Down method) was performed on the PB-201 concentration data simulated by 23 virtual subjects at different doses using Phoenix (v 8.1.0) software to calculate individual PK parameters (including AUC0-24, cmax, t 1/2) after a single dose (day 1) and at steady state (day 7) and to statistically describe the PK parameters (n, mean, standard deviation, coefficient of variation, median, minimum and maximum), and their parameter calculation results were statistically analyzed with PK parameters of chinese phase I clinical study subjects, and the accuracy of model prediction was verified based on the study results.

And further carrying out sensitivity analysis on the covariate ketoconazole which obviously influences PK by combining medicines and foods. Pharmacokinetic profiles (including Cmax, ss, cmin, ss, aucs) were simulated in subjects without co-administration with ketoconazole or with ketoconazole treatment when other covariates were unchanged, and the degree of influence of significant covariates on the pharmacokinetic profiles was quantified. Also, the pharmacokinetic profile (including Cmax, ss, cmin, ss, AUCss) of subjects administered on an empty stomach or with meals was simulated, and the degree of influence of significant covariates on the pharmacokinetic profile was quantitatively described.

Using Phoenix (v 8.1.0) software to carry out NCA analysis on the simulated PB-201 concentration data of 23 virtual subjects under different dosing doses, calculating individual PK parameters (including AUC0-24, cmax and t 1/2) after single dosing (day 1) and at a steady state (day 7), comparing the parameter results calculated by the simulated concentration data with the PK parameters of Chinese phase I clinical study subjects, finding that the PK parameters calculated by the actual measurement data of the Chinese phase I clinical study subjects fall within 5% -95% quantiles of the PK parameters calculated based on the simulated concentration data, and verifying the accuracy of final PK model prediction.

PD simulation mode:

based on the final PopPD model, all patients (n = 23) who simulated the chinese phase I clinical study received the following dosing regimen for 24 weeks without concomitant ketoconazole, morning lunch dosing, in monotherapy and in combination with metformin, respectively: FPG and HbA1c concentrations of PB-201 at 50/50mg, 50/100mg, 100/100mg doses were plotted as mean-time curves for FPG and HbA1c per dose group.

Inter-individual variation (IIV) of PK or PD parameters was described using an exponential model. The residual is described using a proportional error model, the equations are as follows:

P _ij ＝θ _i ×exp(η _ij )

Y _obs ＝Y _pred ×(1+ε1)

wherein, P _ij Refers to the ith parameter value of the jth individual;

θ _i typical values for population parameters;

η _ij refers to the variation between and within the ith parameter of the jth individual, obeying to (0, omega) ₂ ) Normal distribution;

Y _obs and Y _pred The measured concentration value and the predicted concentration value are referred to;

ε 1 refers to the residual, obey (0, σ) ₂ ) A normal distribution.

The PK/PD model of the PB-201 population in healthy people and T2DM patients established in the study is shown in the detailed diagrams in FIGS. 11-13. The simulation view of population PK is shown in FIG. 11; the simulated view of population PD is shown in fig. 12.

And (3) model correction:

for the full-scale model, which only retains very significant covariates, the inter-individual variation and residual error models need to be evaluated. The covariate equations of the model should be as simple as possible, e.g., the power function term parameters reduce to 1 or significantly eliminate the covariate group, which can be redefined with fewer groups or parameters. The optimization fit diagnostic map is also used to test the model.

Evaluation of population PK/PD model

Visual predictive test (VPC) will be used to evaluate the model, using the NONMEM software to simulate the final model 1000 times, comparing the result calculated from the 1000 times of simulation data with the result actually measured; a 5th,50th (median) 95th percentile distribution-time plot of the simulated-based predicted concentrations is plotted, while the measured data and/or the percentile distribution based on the measured data is overlappingly plotted, i.e., a VPC plot, to evaluate the agreement between the model-based data and the measured data. And resampling (Boostrap) the original data set for 500 times, and observing whether the parameters of the final model fall within a 95% confidence interval of the bootstrap parameter result.

Quantitative pharmacological analysis plan

Plan refers to the covariate data of the clinical study subjects in the phase I of China, simulates PK and PD data of a virtual overseas subject population with the same population characteristics, draws a drug concentration time curve and a drug effect curve of blood sugar changing along with the drug concentration, calculates PK/PD non-atrioventricular model parameters, compares the PK/PD non-atrioventricular model parameters with the PK/PD parameters of the clinical study subjects in the phase I of China, and explores ethnic differences. Since the final population PK model only showed that the two demographically uncharacterized covariates, whether co-administered with ketoconazole, and whether administered with food, were significant covariates of the model, ethnicity was not a significant covariate of the final PK model, and was not a significant covariate of the final population PD model. The existing PopPK model constructed based on the overseas population data can better simulate the PK characteristics of different dose groups of PB-201 in Chinese population. In addition, since the non-compartmental model parameters of the subjects PK observed data were statistically calculated in the chinese phase I clinical study, the corresponding PD parameters (fasting glucose, area under the curve (AUEC) for glycated hemoglobin, and peak concentration Cmax were not calculated, so the study only calculated PK non-compartmental model parameters based on the final PK/PD model simulated PB-201 blood concentration data to explore the PD characteristics of PB-201 in T2DM patients during monotherapy and in combination with metformin administration, the study included metformin pool as a covariate of FPG baseline values into the final PD model for population PD simulation (the population PD model with metformin pool added decreased by 1.09 MOFV value compared to the final model).

In order to determine whether the influence of the two screened covariates of ketoconazole combined medication and concomitant food administration on PK has clinical significance, simulation is carried out based on a final population PK model, and the result shows that: when other covariates were unchanged, subjects receiving combined administration of PB201 and ketoconazole had 11.7% increase in AUCss, 6.01% increase in ss, and 22.6% increase in Cmin, ss; when other covariates were unchanged, subjects taking the medication with meals had a 26.3% increase in AUCss, a 38.1% increase in Cmax, and a 10.1% increase in Cmin, ss. Ketoconazole combination and diet have certain effects on PB-201 exposure in healthy subjects and T2DM patients, but further assessment of this effect on exposure-effect (E-R) analysis and PBPK analysis is required to determine whether modifications in the regimen are required for patients co-administered with ketoconazole or with meals. Since the group PD model did not screen covariates with statistically significant significance, no sensitivity analysis of covariates was performed.

Analysis of ethnic differences

The PopPK study included overseas healthy subjects, T2DM patient data, and the impact of ethnicity (1 = caucasian, 2= black, 3= asian, 4= others) on PK exposure was analyzed by exploratory analysis and modeling covariates. Through model covariate analysis, the ethnic factors were not incorporated into the final PopPK model.

The PopPD study was included in outlying T2DM patients, and the impact of ethnicity (1 = caucasian, 2= black, 3= asian, 4= others) on PD markers was analyzed by exploratory analysis and model covariates, and finally no significant impact of ethnicity on PopPD model PD parameters was found.

Conclusion

Analysis of the population pharmacokinetic model showed that a one-compartment model of nonlinear absorption, linear elimination best describes the pharmacokinetic profile of PB-201 in the subject. Results of covariate analysis showed that the effect of ketoconazole in combination with medication and food on exposure of PB-201 in subjects was statistically significant.

The results of the group pharmacodynamic model analysis show that an indirect effect model (promoting FPG elimination) driven linearly by placebo effect and the plasma drug concentration of the central compartment PB-201 best describes the FPG pharmacodynamic characteristics of the subjects after placebo and PB-201. An indirect effect model (promoting HbA1c production) driven linearly with FPG can best describe the HbA1c pharmacodynamic characteristics of a subject following oral administration. No potential covariates were found to significantly affect PD.

Metformin in combination was included as a covariate for the baseline FPG value (FPGbaseline) in the final PD model for population pharmacodynamic model simulations, using the final population PK/PD model, simulating placebo, 50/50mg, 100/50mg and 100/100mg dose groups (administration at breakfast lunch), respectively, with all T2DM patients (n = 23) of the chinese phase I clinical study following 24 weeks of continuous dosing FPG and HbA1c over time on monotherapy and in combination with metformin administration. The results show that, when the T2DM patient receives PB-201 single-drug treatment, after 24 weeks of oral placebo, 50/50mg, 100/50mg and 100/100mg PB-201, the FPG mean value of each dose group is reduced by 0.664mmol/L (50/50 mg), 0.894mmol/L (100/50 mg) and 1.07mmol/L (100/100 mg) respectively after placebo correction compared with the baseline period. Meanwhile, the mean HbA1c of each dose group is reduced by 0.617% (50/50 mg), 0.814% (100/50 mg) and 0.964% (100/100 mg) after placebo correction. In addition, when the patients with T2DM receive PB-201 and metformin, after 24 weeks of oral placebo, 50/50mg, 100/50mg and 100/100mg PB-201, the mean FPG value of each dose group after placebo correction is reduced by 0.652mmol/L (50/50 mg), 0.952mmol/L (100/50 mg) and 1.12mmol/L (100/100 mg) respectively compared with the baseline period. Meanwhile, the mean HbA1c value of each dose group is respectively reduced by 0.622% (50/50 mg), 0.900% (100/50 mg) and 0.980% (100/100 mg) after placebo correction. The results show that PB-201 is continuously reduced during the 24-week simulation period, both on monotherapy and in combination with metformin.

Claims (15)

1. A method for constructing a pharmacokinetic model of a hypoglycemic drug population is characterized by comprising the following steps: firstly, developing a PK model to obtain individual PK parameter values of a subject for constructing a PD model; the PD model is an indirect effect model driven by AUC and is used for describing HbA1c.

2. The construction method according to claim 1, further comprising evaluating covariate factors.

3. The method of claim 1, wherein the PK model is first constructed with a base model and a final model is obtained based on the very significant covariates.

4. The method of claim 3, wherein the PK model base model is:

where Cp refers to the drug concentration in the central compartment.

5. The method of claim 5, wherein the base model of the PD model is:

HbA1c _placebo ＝HbA1c _baseline ·(e ^-HLAM·t )+HSLP·t/168

K _in ＝K _out ·HbA1c _baseline

HbA1c _drug (0)＝0

HbA1c＝HbA1c _drug +HbA1c _placebo

wherein HbA1c _placebo Refers to HbA1c values after placebo administration in patients with T2DM in the middle of the body.

6. The method of claim 1, wherein the hypoglycemic agent is a GLP-1 receptor agonist or glucokinase activator.

7. A pharmacokinetic model of a population of hypoglycemic agents, wherein the hypoglycemic agent is PB-119; the model is a PK-PD model; the PK model is a 1-compartment model of linear elimination and absorption; the PD model is an AUC-driven indirect effect model and is used for describing HbA1c.

8. The population pharmacokinetic model of claim 7, wherein the fundamental PK model is:

where Cp refers to the drug concentration in the central compartment.

9. The population pharmacokinetic model of claim 8, wherein the base model of the PD model is:

HbA1c _placebo ＝HbA1c _baseline ·(e ^-HLAM·t )+HSLP·t/168

K _in ＝K _out ·HbA1c _baseline

HbA1c _drug (0)＝0

HbA1c＝HbA1c _drug +HbA1c _placebo

wherein HbA1c _placebo Refers to HbA1c values after placebo administration in patients with T2DM in the middle of the body.

10. The model of claim 9, further comprising a PK model very significant covariate, being weight and disease progression.

11. A pharmacokinetic model of a hypoglycemic drug population, which is characterized in that the hypoglycemic drug is PB-201; the model is a PK-PD model; the PK model is a 1-chamber model of nonlinear absorption and linear elimination; the PD model is an indirect effect model driven linearly by placebo effect and the plasma drug concentration of the central compartment PB-201.

12. The population pharmacokinetic model of claim 11, wherein the base model of the PK model is

Wherein, V _m Refers to the maximum absorption rate;

K _m refers to the amount of drug that corresponds to half the maximum absorption rate;

A _a refers to the amount of drug in the absorption compartment;

C _p refers to the concentration of drug in the central compartment;

V _c refers to the volume of distribution of the drug in the central compartment;

CL refers to the systemic clearance of the drug; h refers to relative bioavailability;

IMAX refers to the maximum percentage of reduction in bioavailability;

IC50 refers to the dose at which half the maximum percent reduction in bioavailability is achieved;

DOSE refers to the DOSE administered;

plasma concentration-time curves are described for healthy humans and T2DM patients following oral administration of PB-201.

13. The population pharmacokinetic model of claim 12, wherein the base model of the PD model is:

wherein R is _placebo (0)＝1；Kin _placebo ＝Kout _placebo *R _placebo (0)

Wherein R is _drug (0)＝1；Kin _drug ＝Kout _drug *R _drug (0)

ΔFPG _drug ＝FPG _baseline *Fraction*(1-R _drug ) (7)

ΔFPG _placebo ＝FPG _baseline *(1-Fraction)*(1-R _placebo ) (8)

FPG＝FPG _baseline -ΔFPG _drug -ΔFPG _placebo (9)

HbA1c(0)＝HbA1c _baseline (11)

Wherein Placebo refers to a Placebo effect;

kp refers to the rate constant for elimination of placebo effect;

R _placebo refers to the percentage of non-phosphorylated glucose in pancreatic beta-cells and liver driven by placebo effect relative to baseline;

Kin _placebo is denoted by R _placebo A generation rate constant of (a);

Kout _placebo is denoted by R _placebo The elimination rate constant of (2);

R _drug refers to the percentage of unphosphorylated glucose in pancreatic β -cells and liver driven by drug effects relative to baseline;

Kin _drug is denoted by R _drug The generation rate constant of (2);

Kout _drug is denoted by R _drug The elimination rate constant of (2);

k means a reduction in drug concentration R _drug The effect of (c).

14. The population pharmacokinetic model of claim 13, further comprising covariates of the PK model; the covariates are ketoconazole which is used for drug combination, food and sensitivity.

Inter-individual variation (IIV) of pk or PD parameters was described using an exponential model. The residual error is described using a proportional error model, the equations are as follows:

P _ij ＝θ _i ×exp(η _ij )

Y _obs ＝Y _pred ×(1+ε1)

wherein, P _ij Refers to the ith parameter value of the jth individual;

θ _i typical values for population parameters;

η _ij refers to the j (th) individualThe i-th parameter varies between individuals and within individuals and obeys (0, omega) ₂ ) Normal distribution;

Y _obs and Y _pred The measured concentration value and the predicted concentration value are referred to;

ε 1 refers to the residual, obedience (0, σ) ₂ ) A normal distribution.

Images