Palbolizumab population pharmacokinetic model construction method and application thereof

Abstract

The invention relates to a pabolizumab population pharmacokinetic model construction method and application thereof. The method comprises the following steps: collecting structured data at least comprising the concentration of pabolizumab in a blood sample; establishing a basic model of group pharmacokinetics; screening covariables based on the structured data to obtain an optimal covariable; and combining the optimal covariable and the basic model to construct and obtain the group pharmacokinetic model. According to the invention, the optimization and adjustment of the drug dosage can be guided, and the safety and effectiveness of treatment can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pharmacokinetic models, and particularly to a method for constructing a population pharmacokinetic model of pembrolizumab and its application. Background Art

[0002] Pembrolizumab (trade name Keytruda), as a humanized anti-PD-1 inhibitor, has become an important drug in the treatment of various tumors since its launch in 2014. By blocking the interaction between PD-1 and PD-L1 / PD-L2, it has shown significant clinical efficacy in a variety of malignant tumors.

[0003] Although significant progress has been made in the treatment of tumors with pembrolizumab, the research and practice of pharmacokinetic (PK) characteristics still face many challenges, specifically manifested as follows: The metabolism and clearance processes of pembrolizumab in the body are complex, resulting in a high degree of variability in PK characteristics among different patients. This variability makes it difficult for a fixed-dose administration scheme to meet the individualized needs of different patients; a fixed dose may have the problem of overmedication and does not fully consider the individual differences of patients; Although pembrolizumab treatment guided by PK can improve efficacy and control safety, this model has not been widely popularized in clinical practice, and its application is limited by factors such as insufficient data, complexity of model construction, and convenience of clinical implementation; Traditional pharmacokinetic studies require intensive blood sampling from individuals, which is limited by ethics and practical operation difficulties in clinical practice. At the same time, large individual differences are likely to lead to bias in research results, affecting the accuracy and applicability of the model. Summary of the Invention

[0004] The present invention provides a method for constructing a population pharmacokinetic model of pembrolizumab and its application to solve the defects of the prior art.

[0005] The present invention provides a method for constructing a population pharmacokinetic model of pembrolizumab, including:

[0006] S1: Collect structured data including at least the concentration of pembrolizumab in blood samples;

[0007] S2: Establish a basic model of population pharmacokinetics;

[0008] S3: Screen covariates based on the structured data to obtain optimal covariates;

[0009] S4: Combine the optimal covariates and the basic model to construct a population pharmacokinetic model.

[0010] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, wherein the structured data in step S1 includes demographic information, laboratory test results, medication status, and sampling information, and the sampling information includes plasma samples and drug concentrations in blood.

[0011] In one embodiment, the medication status includes single drug (i.e., pembrolizumab) and combination medication; in a preferred embodiment, the combination medication includes: combination medication of pembrolizumab with one or two types of drugs among chemotherapeutic drugs and immunotherapeutic drugs.

[0012] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, step S2 further includes:

[0013] S21: Select a one-compartment model as the structural model of population pharmacokinetics, and the structural model includes an inter-individual variability model and an intra-individual variability model;

[0014] S22: Evaluate the fitting ability of various intra-individual variability models in the structural model, and select the intra-individual variability model with the best fit as the optimal intra-individual variability model;

[0015] S23: Output the inter-individual variability model and the optimal intra-individual variability model as the basic model.

[0016] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, the expression of the inter-individual variability model in step S21 is:

[0017] P i = θ·exp(η i );

[0018] where i is the individual index value, P i is the estimated parameter value of the i-th individual, θ is the population typical value, and η i is the inter-individual difference of the estimated parameter value.

[0019] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, the intra-individual variability models in step S22 include:

[0020] Additive model, and the expression of the additive model is:

[0021] Cobs = C + ε;

[0022] where Cobs is the individual observed value, C is the individual predicted value, and ε is the residual variability;

[0023] Proportional model, and the expression of the proportional model is:

[0024] Cobs = C·(1 + ε);

[0025] Exponential model, the expression of the exponential model is:

[0026] Cobs = C + C power ·ε;

[0027] where power is the relationship parameter between the individual predicted value based on the Poisson distribution and the difference of the parameter residuals;

[0028] Logarithmic transformation additive model, the expression of the logarithmic transformation additive model is:

[0029] ln(Cobs) = ln(C) + ε;

[0030] Hybrid model, the expression of the hybrid model is:

[0031] Cobs = C + ε + C·ε·CMixRatio;

[0032] where CMixRatio is the standard deviation ratio of the proportional error to the additive error.

[0033] According to a method for constructing a pembrolizumab population pharmacokinetic model provided by the present invention, the optimal intra-individual variability model selected in step S22 is the logarithmic transformation additive model.

[0034] According to a method for constructing a pembrolizumab population pharmacokinetic model provided by the present invention, step S3 further includes:

[0035] S31: Define a measure of model fit;

[0036] S32: Add the covariates in the preset first candidate covariate group to the basic model respectively by the forward inclusion method, screen according to the measure, and output the covariates with a measure decrease greater than the first preset threshold as the second candidate covariate group;

[0037] S33: Gradually remove the covariates in the second candidate covariate group from the basic model by the backward elimination method, screen according to the measure, and retain the covariates with a measure increase greater than the second preset threshold or a measure decrease greater than the second preset threshold to obtain a third candidate covariate group;

[0038] S34: Output the covariates in the third candidate covariate group as the optimal covariates.

[0039] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, wherein the first candidate covariate group in step S32 includes: gender, age, weight, body mass index, tumor type, alanine aminotransferase, creatinine, total bilirubin, γ-glutamyl transpeptidase, albumin, total protein, lactate dehydrogenase, glomerular filtration rate, immunotherapy history, surgical history, and medication status;

[0040] The third candidate covariate group in step S34 includes body mass index and albumin.

[0041] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention, wherein the expression of the population pharmacokinetic model constructed in step S4 is:

[0042] V = V tv ·exp(nV);

[0043]

[0044] wherein, V is the central compartment volume of distribution, n is the total number of individuals, V tv is the terminal volume of distribution, nV is the combined volume of distribution of all individuals, CL is the clearance rate, CL tv is the clearance rate corresponding to the terminal volume of distribution, BMI is the body mass index, CL BMI is a parameter characterizing the degree of influence of changes in body mass index on the clearance rate, ALB is the albumin content, CL ALB is a parameter characterizing the degree of influence of changes in albumin on the clearance rate.

[0045] In one embodiment, diagnostic plots and statistical tests are used for internal validation of the model; in a specific embodiment, DV-PRED and DV-IPRED are used to verify the goodness of fit of the model, and the accuracy and stability of the model are evaluated by the bootstrap method.

[0046] The present invention also provides a personalized treatment plan auxiliary decision-making system for solid tumor patients, including:

[0047] A processor;

[0048] The processor receives the body mass index and plasma albumin content data of the to-be-tested solid tumor patient, inputs them into the population pharmacokinetic model constructed by the method for constructing a population pharmacokinetic model of pembrolizumab as described in any one of the above, obtains the clearance rate and volume of distribution, and adjusts the personalized treatment plan of the patient according to the output clearance rate and volume of distribution.

[0049] In one embodiment, the target population is chest solid tumor patients, and the to-be-tested population is chest solid tumor patients using pembrolizumab.

[0050] In one embodiment, the personalized treatment plan includes adjusting the dosage of pembrolizumab; in another embodiment, the personalized treatment plan includes adjusting the type and / or dosage of concomitant medications; in yet another embodiment, the personalized treatment plan includes adjusting the type of concomitant medications. According to the different contents included in the personalized treatment plan, preset personalized treatment plan adjustment rules can be set. When the clearance rate and volume of distribution of pembrolizumab in a patient with solid tumor to be tested are obtained, the adjusted personalized treatment plan for the patient can be obtained according to the preset personalized treatment plan adjustment rules.

[0051] A method for constructing a population pharmacokinetic model of pembrolizumab and its application provided by the present invention uses a one-compartment model based on statistical indicators such as OFV (objective function value), AIC (Akaike information criterion), and BIC (Bayesian information criterion) for a targeted patient population to construct a PPK model of pembrolizumab. This model fills the gap in the field of pharmacokinetic research of pembrolizumab and provides a solid theoretical basis for precision medicine. Secondly, the key pharmacokinetic parameters estimated by the model - clearance rate and volume of distribution - are significantly lower compared to the values in previous research reports, which not only reveals the unique metabolic characteristics of pembrolizumab in specific patients but also provides a scientific basis for subsequent drug dose adjustment and optimization.

[0052] A method for constructing a population pharmacokinetic model of pembrolizumab provided by the present invention aims to collect the blood drug concentration and clinical treatment data of pembrolizumab in patients and establish a PPK model using population pharmacokinetic principles. The present invention comprehensively considers the influence of factors such as patient age, gender, weight, liver and kidney function, tumor type, and treatment plan on the PK characteristics of pembrolizumab. The finally obtained model can provide more accurate individualized dosing recommendations for clinical practice, optimize the treatment effect, and reduce adverse reactions. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0054] Figure 1 It is a schematic flowchart of a method for constructing a population pharmacokinetic model of pembrolizumab provided by an embodiment of the present invention;

[0055] Figure 2 It is a schematic diagram of the goodness of fit between the pembrolizumab basic model provided by an embodiment of the present invention and the population pharmacokinetic model of pembrolizumab of the present invention;

[0056] Figure 3 This is a schematic diagram of the residual distribution of the pembrolizumab population pharmacokinetic model of the present invention;

[0057] Figure 4 This is a schematic diagram of the standardized predicted distribution of the pembrolizumab population pharmacokinetic model of the present invention;

[0058] Figure 5 This is a schematic diagram of the visual predictive check of the pembrolizumab population pharmacokinetic model of the present invention. Detailed implementation manners

[0059] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without making creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention. The following embodiments are used to illustrate the present invention but cannot be used to limit the scope of the present invention.

[0060] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of the present invention. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without conflict, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0061] The following combines with Figures 1 to 5 Describe the embodiments of the present invention.

[0062] As Figure 1 shown, the present invention provides a method for constructing a pembrolizumab population pharmacokinetic model, including:

[0063] S1: Collect structured data including at least the concentration of pembrolizumab in blood samples.

[0064] This invention adopts a prospective research method, collects data of adult patients in the thoracic surgery department of the hospital, and enrolls patients according to inclusion and exclusion criteria. Specific inclusion criteria: patients with solid tumors using pembrolizumab in the thoracic surgery department; gender is not limited, age ≥ 18 years old; specific exclusion criteria: patients whose blood samples were not promptly aliquoted and stored at -80°C; patients with incomplete records of the dosing regimen and sampling time; patients participating in clinical trials.

[0065] For the patients included by the above-mentioned prospective method, the patients are given pembrolizumab at a fixed dose of 200 mg by intravenous infusion over 30 minutes Q3W. Clinicians can adjust the dosing regimen according to the pathophysiological conditions of the patients and determine the treatment duration according to the clinical responses of the patients.

[0066] Before administering pembrolizumab in each cycle, that is, within 60 minutes before each start of infusion, collect plasma samples of the patients, and measure the concentration of pembrolizumab in the serum by a pre-developed and validated ELISA. The average accuracy rate is 84%-100%. The sampling obtains from 1 μg / mL to at least 100 μg / mL. The lower and upper limits of quantification are 3.1 μg / mL and 100 μg / mL respectively, and the coefficient of variation is 5.5%.

[0067] Among them, the structured data in step S1 includes demographic data, laboratory test results, medication conditions, and sampling information. The sampling information includes plasma samples and blood drug concentrations.

[0068] After the blood samples are tested, according to the hospital HIS system, record the basic information, laboratory test results, medication conditions, and sampling information of the patients. The finally obtained information mainly includes: gender, age, height, weight, body mass index, tumor type, white blood cells, lymphocyte percentage, monocyte percentage, neutrophil percentage, absolute lymphocyte count, absolute monocyte count, absolute neutrophil count, red blood cell count, hemoglobin, hematocrit, platelets, alanine aminotransferase, aspartate aminotransferase, γ-glutamyl transpeptidase, alkaline phosphatase, lactate dehydrogenase, a-hydroxybutyric dehydrogenase, creatine kinase, total protein, albumin, total bilirubin, direct bilirubin, albumin / globulin, urea, creatinine, uric acid, estimated glomerular filtration rate, glucose, total cholesterol, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, cortisol, combined medication conditions (including chemotherapy drugs and immunotherapy drugs, 0 means no combined medication, 1 means using chemotherapy drugs, 2 means using immunotherapy drugs), whether there is a surgical history (including surgeries for treating the patient's solid tumors, 0 means no, 1 means yes), and whether there is an immunotherapy history (including drugs such as nivolumab, 0 means no, 1 means yes).

[0069] S2: Establish a basic model of population pharmacokinetics.

[0070] The method for estimating the model parameters of the present invention adopts the first-order conditional estimation - extended least squares method (FOCE-ELS), and tests the one-compartment, two-compartment, and three-compartment models with zero-order absorption as the pharmacokinetic structural models of pembrolizumab. The structural models are compared according to the values of the objective function value (OFV), Akaike information criterion (AIC), and Bayesian information criterion (BIC), the goodness-of-fit graph, and the precision of the estimated values of THETA, OMEGA, and SIGMA.

[0071] Among them, step S2 further includes:

[0072] S21: Select a one-compartment model as the structural model of population pharmacokinetics, and the structural model includes the between-subject variability model and the within-subject variability model.

[0073] In PPK analysis, the structural model defines the basic form of the change in drug concentration over time. For pembrolizumab studied in the present invention, its random effects include between-subject variability (BSV) and within-subject variability (WSV). Generally, it is assumed that both between-subject variability and within-subject variability conform to the normal distribution. Therefore, the following exponential model can be obtained for between-subject variability, where the mean of the between-subject differences in the estimated parameter values is 0 and the variance is σ 2 of the normal distribution.

[0074] Among them, the expression of the between-subject variability model in step S21 is:

[0075] P i = θ·exp(η i );

[0076] Among them, i is the individual index value, P i is the estimated parameter value of the i-th individual, θ is the population typical value, and η i is the between-subject difference in the estimated parameter value.

[0077] S22: Evaluate the fitting ability of various within-subject variability models in the structural model, and select the within-subject variability model with the best fit as the optimal within-subject variability model.

[0078] The within-subject variability is respectively fitted with additive, proportional, exponential, log-transformed additive, and mixed models in sequence, and the residual variability model with the best goodness-of-fit is selected.

[0079] Among them, the within-subject variability models in step S22 include:

[0080] Additive model, and the expression of the additive model is:

[0081] Cobs = C + ε;

[0082] Among them, Cobs is the individual observation value, C is the individual predicted value, and ε is the residual variation difference;

[0083] Proportional model, the expression of the proportional model is:

[0084] Cobs = C·(1 + ε);

[0085] Exponential model, the expression of the exponential model is:

[0086] Cobs = C + C power ·ε;

[0087] Among them, power is the relationship parameter between the individual predicted value based on the Poisson distribution and the residual variation difference;

[0088] Logarithmic transformation additive model, the expression of the logarithmic transformation additive model is:

[0089] ln(Cobs) = ln(C) + ε;

[0090] Mixed model, the expression of the mixed model is:

[0091] Cobs = C + ε + C·ε·CMixRatio;

[0092] Among them, CMixRatio is the standard deviation ratio of the proportional error to the additive error.

[0093] Among them, the optimal within-individual variability model selected in step S22 is the logarithmic transformation additive model.

[0094] S23: Output the between-individual variability model and the optimal within-individual variability model as the basic model.

[0095] In step S3, since the one-compartment model has a smaller OFV, AIC, BIC and better goodness of fit compared with the two-compartment model and the three-compartment model, and is more suitable for describing the plasma drug concentration data of the present invention, the one-compartment model is first selected in step S31. Subsequently, in step S22, five within-individual variability models of additive, proportional, exponential, logarithmic transformation additive and mixed types are investigated respectively in the present invention, and finally the logarithmic transformation additive type is selected to describe the within-individual variability.

[0096] S3: Screen covariates based on the structured data to obtain the optimal covariates.

[0097] In step S3, the covariates to be investigated are mainly subjected to covariate correlation analysis to eliminate collinearity and instability of the parameter estimation values, which is used for the establishment of the pembrolizumab PPK model in step S4.

[0098] Among them, step S3 further includes:

[0099] S31: Define a measure of the goodness of fit of the model.

[0100] When screening covariates, the present invention first uses a graphical method to preliminarily judge the correlation between covariates and model variation, and then defines the OFV as -2 times the log-likelihood value (-2LL), as an overall measure of the goodness of fit. Subsequently, in step S32, a stepwise method of forward inclusion and backward elimination is used to screen covariates in the pembrolizumab modeling process.

[0101] S32: Add the covariates within the preset first candidate covariate group to the basic model respectively by the forward inclusion method, screen according to the measure, and output the covariates with a decrease in the measure greater than the first preset threshold as the second candidate covariate group.

[0102] In the forward inclusion process of step S32, first add covariates to the basic model one by one, add only one covariate each time, and only try on one parameter. Assuming the significance level of the test α = 0.01, if the OFV drops by more than 6.64 after adding a certain covariate, then add this covariate to the model, otherwise eliminate it.

[0103] Among them, the first candidate covariate group in step S32 includes: gender, age, weight, body mass index, tumor type, alanine aminotransferase, creatinine, total bilirubin, γ-glutamyl transpeptidase, albumin, total protein, lactate dehydrogenase, glomerular filtration rate, immunotherapy history, surgical history, medication status.

[0104] S33: Gradually eliminate the covariates within the second candidate covariate group from the basic model by the backward elimination method, screen according to the measure, and retain the covariates with an increase in the measure greater than the second preset threshold or a decrease greater than the second preset threshold to obtain the third candidate covariate group.

[0105] Among them, the measure in step S31 is defined as the objective function value of -2LL, the first preset threshold in step S32 is 6.64, and the second preset threshold in step S33 is 10.83.

[0106] In the backward elimination process of step S33, only one covariate is eliminated in each round. Assuming the significance level of the test α = 0.001, that is, if the change in OFV is greater than 10.83, then it is considered that this factor has significant significance and is retained in the model. After excluding the fixed effect parameters without significant significance through the backward elimination process, the final model is obtained.

[0107] S34: Output the covariates in the third candidate covariate group as the optimal covariates.

[0108] Among them, the third candidate covariate group in step S34 includes body mass index and albumin.

[0109] S4: Construct a population pharmacokinetic model by combining the covariates and the basic model.

[0110] Among them, the expression of the population pharmacokinetic model constructed in step S4 is:

[0111] V = V tv ·exp(nV);

[0112]

[0113] Among them, V is the central compartment volume of distribution, n is the total number of individuals, V tv is the terminal volume of distribution, nV is the combined volume of distribution of all individuals, CL is the clearance rate, CL tv is the clearance rate corresponding to the terminal volume of distribution, BMI is the body mass index, CL BMI is a parameter representing the degree of influence of the change in body mass index on the clearance rate, ALB is the albumin content, CL ALB is a parameter representing the degree of influence of the change in albumin on the clearance rate.

[0114] After obtaining the model established in step S4, the present invention uses an internal validation method combining diagnostic plots and statistical tests to evaluate the PPK model of pembrolizumab, and objectively evaluates the predictive ability of the model within the target range and whether the model defects will affect the decision-making.

[0115] The GOF plot can reflect the goodness of fit of the model to the observed data and the prediction bias of the model, including the scatter plots of observed values (DV) - population predicted values (PRED) of the basic model and the final model, DV - IPRED scatter plots, conditional weighted residuals (CWERS) - time after dosing (TAD) scatter plots, CWERS - PRED scatter plots, weighted residual histograms and Q - Q plots. The points of the DV - PRED and DV - IPRED scatter plots tend to the reference line (y = x), and the CWERS is symmetrically distributed on both sides of the reference line (y = 0) and most are within ±2, indicating that the model fits well.

[0116] The accuracy and stability of the model are evaluated by the bootstrap method. The bootstrap sampling is repeated 1000 times to generate a new set of datasets. The final model to be evaluated and the bootstrap datasets are respectively used for parameter fitting to estimate the model parameters. Calculate the proportion of the model successfully estimating the parameters, that is, the model robustness rate information. For example, if the model to be evaluated is successful 900 times in 1000 sets of bootstrap datasets, it means the robustness rate is 90%. Summarize the calculation results of the successful estimations, calculate the median and 95% confidence interval (2.5% - 97.5%) of the parameters, and compare them with the original estimated values of the final model parameters. If the 2.5% - 97.5% interval of each parameter contains the original model estimated parameters and meets the pre-set robustness rate, the model is considered to have good stability.

[0117] Use VPC to generate model-based simulated datasets, and evaluate the performance of the final model by statistical methods to evaluate the accuracy and predictive ability of the model. In the present invention, the internal dataset is simulated 1000 times, and the 5%, 50%, and 95% quantiles of the observed dataset and the simulated dataset at each time point are calculated respectively. If the number of observed values outside the 90% prediction interval does not exceed 10%, it is considered that the model has good predictive efficacy.

[0118] The present invention also provides an auxiliary decision-making system for personalized treatment plans for solid tumor patients, including:

[0119] At least one processor;

[0120] The processor receives the body mass index and plasma albumin content data of the solid tumor patient to be tested, and inputs them into the population pharmacokinetic model constructed by the pembrolizumab population pharmacokinetic model construction method described in any one of the above, obtains the clearance rate and volume of distribution, and adjusts the personalized treatment plan of the patient according to the output clearance rate and volume of distribution.

[0121] Specifically, the personalized treatment plan includes adjusting the dose of pembrolizumab. The personalized treatment plan may also include adjusting the types and / or doses of concomitant medications, and the personalized treatment plan may also include adjusting the types of concomitant medications.

[0122] The following describes a pembrolizumab population pharmacokinetic model construction method provided by the present invention in combination with a specific embodiment.

[0123] During the preset sampling time of 6 months, a total of 46 patients with chest solid tumors in the hospital's thoracic surgery department were included. Among them, there were 9 female patients (80.4%) and 37 male patients (19.6%). The average age was 62.6 years old, and the average weight was 69.4 kg. The tumor types included 36 cases of lung cancer (78.3%), 9 cases of esophageal cancer (19.6%), and 1 case of mesothelioma (2.2%). The demographic data and clinical indicators are shown in Tables 1 and 2 respectively. Finally, 89 effectively monitored plasma samples were collected, all of which were trough concentration points, and 1 - 4 samples were taken from each patient. The plasma sample concentration was measured by ELISA method in the range of 13.2 - 104.0 μg / mL.

[0124] Table 1 Demographic data of the dataset in this example

[0125] Categorical variable n(％) Gender Male 37(80.4) Female 9(19.6) Tumor type Lung cancer 36(78.3) Esophageal cancer 9(19.6) Mesothelioma 1(2.2) Previous immunotherapy Yes 10(21.7) No 36(78.3) Surgical history Yes 11(23.9) No 35(76.1)

[0126] Table 2 Clinical indicators of the dataset in this example

[0127] Continuous variable Mean (standard deviation) Median (range) Age (years) 62.6(8.7) 64(58-68) Weight (kg) 69.4(13.6) 69(60-76) <![CDATA[Body mass index (kg / m 2 )]]> 24(4.1) 23.3(20.9-26.7) Creatinine (μmol / L) 70(18.8) 67(56.5-75.5) Total protein (g / L) 67.5(6.3) 66.8(62.7-70.7) Albumin (g / L) 38.9(3.6) 38.9(36.2-41.9) Alanine aminotransferase (U / L) 22.3(14.6) 19(14-27) Total bilirubin (μmol / L) 9.3(3.4) 8.8(7-11) <![CDATA[Glomerular filtration rate (mL / min / 1.73m 2 )]]> 92.2(15.9) 95.9(87.1-102) Lactate dehydrogenase (U / L) 186.2(46.2) 180(158.5-202)

[0128] Compared with the two - compartment model and the three - compartment model, the one - compartment model has a smaller OFV, AIC, BIC and better goodness of fit, and is more suitable for describing the plasma drug concentration data of the present invention. The present invention investigated five within - subject variability models: additive, proportional, exponential, log - transformed additive, and mixed type. Finally, the log - transformed additive type was selected to describe the within - subject variability. The basic model parameters of pembrolizumab obtained in this example are shown in Table 3. In Table 3, RSE is the relative standard error, CL is the clearance rate, V is the central compartment volume of distribution, ω 2 -CL is the between - subject variability of pembrolizumab clearance rate, and Log - additive is the log - transformed additive type residual.

[0129] Table 3 Estimation results of the basic model parameters of pembrolizumab in this example

[0130] Parameter Estimate RSE (%) 95% CI (lower) 95% CI (upper) PK model parameter CL (mL / hr) 3.61 6.7 3.13 4.09 V (L) 1.78 13.0 1.32 2.23 Interindividual variability <![CDATA[ω 2 -CL]]> 0.057 24.2 - - Residual model Log-additive 0.48 6.7 0.42 0.55

[0131] After establishing the basic model, covariates were included in the investigation to screen out the covariates that have an important impact on the pharmacokinetic parameters of pembrolizumab, and a covariate model was constructed. The screening results by the Stepwise method of forward inclusion (P < 0.01) and backward elimination (P < 0.001) showed that BMI and ALB had a significant impact on the clearance rate of pembrolizumab in patients with chest solid tumors. BMI and ALB were included as covariates in the final model. The final model is shown in the model expression in the aforementioned step S4 and will not be elaborated here. The final model parameters are shown in Table 4.

[0132] Table 4 Estimation results of the final model parameters of pembrolizumab in this example

[0133] Parameter Estimate RSE (%) 95% CI (lower) 95% CI (upper) PK model parameter CL (mL / hr) 3.59 5.17 3.22 3.96 V (L) 1.93 14.7 1.37 2.50 Covariate CL-BMI 0.98 25.2 0.49 1.48 CL-ALB -2.34 26.4 -3.58 -1.12 Interindividual variability <![CDATA[ω 2 -CL]]> 0.029 17.0 0.016 0.075 Residual model Log-additive 0.35 9.7 0.28 0.42

[0134] The goodness-of-fit (GOF) plots of the basic model and the final model were separately drawn to evaluate the fitting effect and prediction deviation of the model, as shown in Figure 2 and Figure 3 . The relative tightness of the scatter plots of the population predicted concentration values, individual predicted concentration values, and observed concentration values of the final model clustering near y = x was better than that of the basic model, indicating that the goodness-of-fit of the final model was better than that of the basic model. For the scatter plot of the conditional weighted residuals against the time after the previous dose and the population predicted values, the conditional weighted residuals of the final model were mostly evenly distributed between y = ±2, and the overall trend of its data fitting was better than that of the basic model, indicating that the fitting effect of the final model was good. The bar chart and Q-Q plot of the normal test of the conditional weighted residuals of the final model showed that the conditional weighted residuals had basically no deviation and were normally distributed, as shown in Figure 4 .

[0135] Figure 2 In Figure 2 , (A) shows the observed blood drug concentration values of the basic model and the final model, and Figure 2 , (B) shows the scatter plot of the individual predicted values of the basic model and the final model; Figure 2 , (C) shows the observed blood drug concentration values of the basic model and the final model, Figure 2 , (D) shows the scatter plot of the population predicted values of the basic model and the final model; Figure 3 In Figure 3 , (A) shows the scatter plot of the conditional weighted residuals / time after dosing, Figure 3 , (B) shows the scatter plot of the individual weighted residuals / individual predicted values; Figure 4 In Figure 4 , (A) shows the bar chart of the standardized prediction error distribution, Figure 4 , (B) shows the standard normal distribution Q-Q plot of the conditional weighted residuals.

[0136] In the present invention, the bootstrap method was used to internally validate the final model to investigate the stability of the model. A total of 1000 sets of bootstrap data sets were generated, and the number of successful validations was 1000 times, and the model robustness rate was 100%. The estimated values of each parameter of the final model were close to the median of the data generated by the bootstrap analysis (|deviation| < 15%), and the estimated values of each parameter were within the 95% confidence interval of the parameter estimated values of the bootstrap method. The model had good stability. The final model of pembrolizumab and the parameter estimation results of the bootstrap method are shown in Table 5.

[0137] Table 5 Parameter Estimation Results of the Final Model of Pembrolizumab in This Example and the Bootstrap Method

[0138]

[0139]

[0140] Visual predictive check (VPC) plots were made for the final model to visually evaluate the accuracy of the model's predictions. Since all points fell within the 90% prediction interval, the final model was considered to have good predictability, as shown in Figure 5 .

[0141] The present invention established a pembrolizumab PPK model, which laid a solid foundation for the implementation of precision medicine. Using a one-compartment model, this model accurately estimated the pharmacokinetic parameters specific to a particular patient (significantly reduced CL and V). This new finding helps guide the optimization of drug dosage, improve the safety and effectiveness of treatment, and enables clinicians to implement more individualized dosing regimens, effectively reducing adverse reactions and improving the quality of life of patients. In addition, the successful practice of the present invention not only promotes the precise application of pembrolizumab, but also provides a reference for population pharmacokinetic studies of anti-tumor drugs, accelerating the popularization and development of precision medicine in the field of tumor treatment.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. However, such modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for constructing a population pharmacokinetic model of pembrolizumab, characterized in that: include: S1: Collect structured data including at least the concentration of pembrolizumab in blood samples; S2: Establish a basic model for population pharmacokinetics; S3: screening covariates based on the structured data to obtain optimal covariates; S4: Combining the optimal covariate with the basic model to construct a population pharmacokinetic model.

2. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 1, characterized in that: The structured data in step S1 includes demographic information, laboratory test results, medication status, and sampling information, and the sampling information includes plasma samples and blood drug concentrations.

3. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 1, characterized in that: Step S2 further comprises: S21: selecting a one-compartment model as a structural model of population pharmacokinetics, wherein the structural model includes an inter-individual variation model and an intra-individual variation model; S22: evaluating the fitting ability of multiple intra-individual variation models in the structural model, and selecting the intra-individual variation model with the best fitting as the optimal intra-individual variation model; S23: Outputting the inter-individual variation model and the optimal intra-individual variation model as the basic model.

4. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 3, characterized in that: The expression of the inter-individual variation model in step S21 is: P i =θ·exp(η i ); Among them, i is the individual index value, P i is the estimated parameter value of the ith individual, θ is the typical value of the population, η i To estimate the inter-individual differences in parameter values.

5. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 3, characterized in that: The intra-individual variation model in step S22 includes: Additive model, the expression of the additive model is: Cobs = C + ε; Among them, Cobs is the individual observation value, C is the individual prediction value, and ε is the parameter residual difference; A proportional model, wherein the expression of the proportional model is: Cobs = C·(1+ε); Exponential model, the expression of the exponential model is: Cobs=C+C power ·e; Among them, power is the relationship parameter between the individual predicted value and the residual variation difference based on Poisson distribution; Logarithmically transformed additive model, the expression of the logarithmically transformed additive model is: ln(Cobs)=ln(C)+ε; A hybrid model, wherein the expression of the hybrid model is: Cobs=C+ε+C·ε·CMixRatio; Where CMixRatio is the ratio of the standard deviation of the proportional error to the additive error.

6. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 5, characterized in that: The optimal intra-individual variation model selected in step S22 is the logarithmic transformation additive model.

7. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 1, characterized in that: Step S3 further comprises: S31: Define the measure of model fit; S32: adding the covariates in the preset first candidate covariate group to the basic model respectively by the forward inclusion method, screening them according to the measurement index, and outputting the covariates whose measurement index drops by more than the first preset threshold as the second candidate covariate group; S33: gradually eliminating the covariates in the second candidate covariate group from the basic model by a reverse elimination method, screening according to the measurement index, retaining the covariates whose measurement index increases by more than a second preset threshold or decreases by more than the second preset threshold, to obtain a third candidate covariate group; S34: Outputting the third candidate covariate group as the optimal covariate.

8. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 7, characterized in that: The first candidate covariate group in step S32 includes: gender, age, weight, body mass index, tumor type, alanine aminotransferase, creatinine, total bilirubin, γ-glutamyl transpeptidase, albumin, total protein, lactate dehydrogenase, glomerular filtration rate, immunotherapy history, surgical history, and medication status; The third candidate covariate group in step S34 includes body mass index and albumin.

9. The method for constructing a population pharmacokinetic model of pembrolizumab according to claim 8, characterized in that: The expression of the population pharmacokinetic model constructed in step S4 is: V=V tv ·exp(nV); Where V is the central compartment distribution volume, n is the total number of individuals, V tv is the terminal distribution volume, nV is the comprehensive distribution volume of all individuals, CL is the clearance rate, CL tv is the clearance corresponding to the terminal distribution volume, BMI is the body mass index, CL BMI ALB is the albumin content, CL is the effect of body mass index on clearance. ALB It is a parameter that characterizes the impact of albumin changes on clearance.

10. A personalized treatment plan decision support system for patients with solid tumors, characterized in that: include: processor; The processor receives body mass index and plasma albumin content data of the solid tumor patient to be tested, and inputs the data into the population pharmacokinetic model constructed by the method for constructing a population pharmacokinetic model of pembrolizumab as described in any one of claims 1 to 9, obtains the clearance rate and distribution volume, and adjusts the patient's personalized treatment plan according to the output clearance rate and distribution volume.

Images