WO2023241402A1 - 药物吸收速率常数的预测模型、设备和存储介质 - Google Patents

药物吸收速率常数的预测模型、设备和存储介质 Download PDF

Info

Publication number
WO2023241402A1
WO2023241402A1 PCT/CN2023/098532 CN2023098532W WO2023241402A1 WO 2023241402 A1 WO2023241402 A1 WO 2023241402A1 CN 2023098532 W CN2023098532 W CN 2023098532W WO 2023241402 A1 WO2023241402 A1 WO 2023241402A1
Authority
WO
WIPO (PCT)
Prior art keywords
value
drug
model
tablets
absorption
Prior art date
Application number
PCT/CN2023/098532
Other languages
English (en)
French (fr)
Inventor
易木林
Original Assignee
湖南慧泽生物医药科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 湖南慧泽生物医药科技有限公司 filed Critical 湖南慧泽生物医药科技有限公司
Publication of WO2023241402A1 publication Critical patent/WO2023241402A1/zh

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H70/00—ICT specially adapted for the handling or processing of medical references
    • G16H70/40—ICT specially adapted for the handling or processing of medical references relating to drugs, e.g. their side effects or intended usage
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/50—Molecular design, e.g. of drugs

Definitions

  • the invention relates to the field of drug information technology, and in particular to a prediction model, equipment and storage medium for drug absorption rate constants.
  • Drug absorption refers to the process of drug uptake from the administration site into the blood circulation.
  • the absorption rate of drugs is an important parameter in pharmacokinetics and bioavailability studies.
  • the rate of absorption has a great influence on blood concentration and is affected by the route of administration and many other factors. Different routes of administration naturally lead to different absorption rates; different dosage forms also lead to different absorption rates.
  • the oral absorption of solid dosage forms depends on the disintegration of the preparation, the dissolution of the active drug, the drug concentration and blood circulation at the absorption site, and the location and area of the absorption surface.
  • the drug absorption rate constant ka value is a relative index indicating the rate at which the drug enters the blood system at the absorption site. It is also a major model parameter in the absorption pharmacokinetics model.
  • the k a value also plays a leading role in the calculation of the peak time t max reached by the drug in the body and the peak concentration C max of the drug in the body, and in the determination of the dosage regimen, the calculation of the sustained release rate of buffered long-acting agents It also has a relatively important meaning.
  • Commonly used quantitative methods for estimating drug absorption rate constant Ka include compartmental model methods (Wagner-Nelson method, Loo-Riegelman method) and non-compartmental model methods (deconvolution integral method and statistical moment method).
  • the accuracy of estimating k a by the Wagner-Nelson method and the Loo-Riegelman method is significantly higher than that of the non-compartmental model method, but the parameters of the compartment model need to be accurately analyzed, and intravenous data for multiple drugs are lacking. As a result, Loo-Riegelman cannot be applied, etc.
  • the purpose of the present invention is to provide a prediction model, equipment and storage medium for drug absorption rate constants, and to estimate the ka value by constructing an absorption kinetic model that does not depend on the compartment model.
  • a compartmental model to obtain the real k a value, and the difference is verified by estimating the k a value and the real k a value, and the accuracy of the k a value is further verified based on the pharmacokinetic parameters.
  • the k a value predicted by the prediction model of the present invention is accurate. It is highly accurate and does not require venous blood drug concentration data. It can estimate the drug absorption rate constant ka value of different types of drugs, thereby providing an important tool for the analysis of in vivo absorption kinetics of pharmaceutical preparations and the in vivo evaluation of IVIVC.
  • the present invention provides a prediction model of drug absorption rate constant, the model includes:
  • the acquisition module is used to obtain the measured blood drug concentration and time sampling points of the drug preparation, and then draw the measured drug time curve to obtain the pharmacokinetic parameters of the drug's in vivo absorption characteristics (such as C max , T max , AUC 0-t , AUC 0 - ⁇ and elimination half-life (t 1/2 ));
  • the fitting module outputs an absorption kinetic model that does not depend on the compartment model based on the measured drug time curve and T max , and obtains the estimated k a value;
  • the real module selects the compartment model pre-input function, measured blood drug concentration, time sampling point, and fixed parameters according to the blood drug concentration characteristics to obtain the real ka value;
  • the comparison module compares the difference between the estimated ka value and the real ka value, and outputs an estimated ka value with high accuracy.
  • a correction module which uses the estimated ka value to predict the absorption phase of the pharmaceutical preparation and the predicted C max value, compares the true C max and the predicted C max , and outputs the best estimated ka value.
  • the measured drug time curve is divided into two processes: linear kinetic increase and linear kinetic decrease of blood drug concentration over time.
  • the absorption phase of the measured drug time curve is carried out.
  • the deconvolution integral is used to obtain the drug absorption curve in vivo, and the expression relationship between blood drug concentration and time Ct is constructed, thereby establishing an absorption kinetic model that does not depend on the compartment model, in which the Ct expression relationship is:
  • k a is the linear dynamics rising rate constant
  • k is the linear dynamics falling rate constant
  • A is the correction coefficient
  • Formula 1 constructed by the present invention is a C-t relational expression of an in vivo absorption kinetic model that does not depend on the compartment model and is only related to the drug time curve characteristics.
  • Step 1 Based on the actual measured blood concentration of the drug preparation, after logarithmizing the blood drug concentration C, take several consecutive values after reaching the peak and fit a linear regression.
  • Step 2 Based on formula 1 and the A value obtained in step 1, use the Python iteration method program to set the k value and k a value according to the set value range (for example, k a is 0.01 ⁇ 10.0h -1 , k is 0.01 ⁇ 5.0h -1 ) Continuously traverse into Formula 1, and iterate every 0.01 value to obtain the predicted drug time curve that minimizes the sum of absolute values of the residual values of the actual measured drug time curve;
  • Step 3 Use Formula 3 to calculate the minimum sum of the absolute values of the residual values of the estimated blood concentration and the measured blood concentration of the drug preparation in the drug-time curve obtained in Step 2, and then output the optimal ka value;
  • C i is the measured blood drug concentration
  • C i' is the estimated blood drug concentration. The smaller the SUM value, the better the fit.
  • the optimal ka value is the estimated ka value of an absorption kinetic model in which in vivo absorption does not depend on the compartment model.
  • k a is the linear kinetics rising rate constant, that is, the drug absorption rate constant; k is the linear kinetics falling rate constant, that is, the elimination rate constant.
  • WinNonlin software version 8.2, Certara Company
  • k a is the linear kinetics rising rate constant, that is, the drug absorption rate constant;
  • k is the linear kinetics falling rate constant, that is, the elimination rate constant;
  • k 12 refers to the drug from the central chamber (blood) to the peripheral chamber in the two-chamber model
  • the rate constant of (organ, tissue) diffusion k 21 refers to the rate constant of drug diffusion from the peripheral chamber (organ, tissue) to the central chamber (blood) in the two-compartment model,
  • k 10 refers to the rate constant of the drug from the central chamber in the two-compartment model (blood) elimination rate constant.
  • ⁇ represents the first-order rate constant of distributed phase mixing
  • ⁇ represents the first-order rate constant of elimination phase mixing, which are calculated by Formula 7 and Formula 8 respectively;
  • the drug time curve is calculated using the model of the present invention.
  • the k a value calculated by the model is basically consistent with the real k a value.
  • the drugs that meet the dual-compartment model are abiraterone acetate tablets, acyclovir suspension, azithromycin tablets, benazepril capsules, bupropion tablets, candesartan medoxomil tablets, captopril Li tablets, celecoxib capsules, ciprofloxacin tablets, clopidogrel tablets, daclatasvir tablets, domperidone tablets, drotaverine tablets, glyburide tablets, hydrochlorothiazide tablets, isradipine capsules, Triconazole tablets, lacidipine tablets, lercanidipine hydrochloride tablets, levonorgestrel tablets, loratadine tablets, metformin tablets, mycophenolate mofetil tablets, naproxen tablets, olmesartan medoxomil tablets, Oseltamivir phosphate capsules, quinapril tablets, repaglinide tablets, rilpivirine tablets, rosuvastatin tablets
  • k a is 1.098, 0.603, 0.375h -1 ; k 12 is 0.525, 0.211, 0.133h -1 ; k 21 is 0.176, 0.067, 0.025h -1 ; k 10 is 0.571, 0.271 and 0.100h -1 ;According to the relationship between k a , k 12 , k 21 , k 10 (that is, k a >k 12 +k 10 , and ka >k 12 >k 21 ), randomly combine k a , k 12 , k 21 and k a value of 10 , and calculate the drug-time curve of each group according to formulas 6 to 8; compared with the real k a value, the estimated k a value using an absorption kinetic model that does not depend on the compartment model has a positive value for RE and Negative values, all RE values are within ⁇ 16%, and most of them are within ⁇ 10%, indicating that the accuracy of the ka values analyzed by
  • the results demonstrate that the parameters (k 12 , k 21 , k 10 , etc.) of the absorption kinetics model that are independent of the compartment model, without venous blood drug concentration data, estimate k from the absorption kinetics model that is independent of the compartment model.
  • the a value has high accuracy and meets the k a value analysis of different types of drugs.
  • the pharmaceutical preparation is carbamazepine tablets or cyclosporine soft capsules.
  • the present invention provides a method for verifying the accuracy of drug absorption rate constants.
  • an actual measured drug time curve is drawn; based on the actual measured drug time curve, an absorption kinetic model that does not depend on the compartment model is established. , calculate the absorption rate constant k a of the corresponding model; verify the accuracy of the absorption rate constant k a calculated by the absorption kinetic model that does not depend on the compartment model based on the existing compartment model parameter settings and clinical trial data.
  • the verification method includes the following steps:
  • k a is the linear kinetic rising rate constant (i.e., the absorption rate constant of the drug); k is the linear kinetic falling rate constant; A is the correction coefficient;
  • S6 Perform Pearson correlation analysis (SPSS 25.0) on the pharmacokinetic parameters (C max , T max , C max /AUC 0-t and other pharmacokinetic parameters that reflect the absorption characteristics in the body) and the estimated k a value. , SPSS Inc.), to further verify the accuracy of the k a value.
  • SPSS 25.0 Pearson correlation analysis
  • the estimated C max is calculated by using the estimated ka value.
  • the estimated k a value has a good correlation with the pharmacokinetic parameters (C max , T max, etc.) that reflect the absorption characteristics of the drug in the body, and can also accurately predict the absorption phase and C max value of the drug.
  • the present invention provides a method for obtaining a drug absorption rate constant, which includes the following steps:
  • k a is the linear kinetic rising rate constant (i.e., the absorption rate constant of the drug); k is the linear kinetic falling rate constant; A is the correction coefficient;
  • Estimated k a value based on the iterative method to analyze the absorption kinetic model that does not depend on the compartment model.
  • the present invention provides a method for predicting the C max of a pharmaceutical preparation using the absorption kinetic model established in the present invention that is independent of the compartment model.
  • the steps are as follows:
  • the present invention provides a computer device, including a memory and a processor.
  • the memory stores a computer program.
  • the execution of the computer program by the processor is a step to implement the verification method of the present invention.
  • the present invention provides a computer-readable storage medium that stores a computer program, and the computer program is executed by a processor to implement the steps of the verification method of the present invention.
  • the present invention comprehensively considers the blood concentration of the drug, sampling time point, single dose, total drug amount, C max , T max , AUC 0-t , AUC 0- ⁇ , elimination half-life (t 1/2 ) and other constructions.
  • Key pharmacokinetic parameters were modeled to construct an absorption kinetic model that was independent of the compartment model.
  • the estimated k a value obtained by the absorption kinetic model, Wagner-Nelson method, Loo-Riegelman method, deconvolution integral method and statistical moment method was compared with the real k a value obtained by the chamber model setting. The method verifies the accuracy of the estimated k a value to accurately reflect the absorption phase and C max of the drug in the body.
  • the absorption kinetic model constructed by the method of the present invention that does not rely on the compartment model solves the shortcomings of the traditional method of estimating the k a value, and creatively proposes an in vivo absorption kinetic model that does not rely on the compartment model and is only related to the drug time curve characteristics.
  • Formula 1 of the learning model is used to improve the accuracy of analytically estimating the ka value .
  • the k a value analyzed by the drug time curve prediction model of the pharmaceutical preparation provided by the present invention is highly accurate, has a wide range of applications, and can provide analysis of the in vivo absorption kinetics of the pharmaceutical preparation and in vivo evaluation of its in vivo and in vitro correlation (IVIVC). Important tool.
  • the method of the present invention was successfully applied to the ka analysis of two model drugs, carbamazepine tablets and cyclosporine soft capsules, and was compared with the pharmacokinetic parameters (T max , C max , C max /AUC) that reflect the absorption characteristics of the drug in the body. 0-t, etc.) have good correlation, can also accurately predict the absorption phase and C max of the drug, and can also be used for quality control of reference preparations and test preparations.
  • Figure 1 Accuracy of the absorption kinetic model independent of the compartment model and the Wagner-Nelson method for estimating the k a value of a single-compartment model drug.
  • Figure 2 Drug time curve of the dual-compartment model parameter setting group (39 sets of data in total).
  • Figure 3 Human body drug time curve after oral administration of carbamazepine tablets and cyclosporine soft capsules.
  • Figure 4 Measured and predicted average drug duration curves of pharmaceutical preparations.
  • Carbamazepine tablets A) fasting - reference preparation, (B) fasting - test preparation, (C) postprandial - reference preparation, (D) postprandial - test preparation; cyclosporine soft capsules ( E) Fasting - reference preparation, (F) Fasting - test preparation, (G) Postprandial - reference preparation, (H) Postprandial - test preparation.
  • the drug is administered as a single dose X 0 , the total amount of drug reaching the absorption site is X a .
  • the drug is absorbed in a first-order rate process (k a ), and the amount of drug entering the human body is ) are distributed (k 12 ) and eliminated (k 10 ) to the peripheral chambers (organs, tissues) in a first-order rate process, and the drugs in the peripheral chambers also return to the central chamber (k 21 ) in a first-order rate process.
  • the drug-time curve is bounded by T max and can be divided into an absorption phase and a disposal phase (the single-compartment model is equivalent to the elimination phase; the dual-compartment model is the sum of the distribution phase and the elimination phase).
  • the absorption phase the absorption rate of the drug is always greater than the disposal rate, causing the blood drug concentration to continue to rise; when T max is reached, the drug absorption rate is equal to the disposal rate; thereafter, distribution and/or elimination will dominate, causing the blood drug concentration to continue to rise. Drop until elimination is complete.
  • the two absorption curves After integrating the absorption inverse convolutions of the characteristic drug-time curves of single-compartment model and dual-compartment model drugs (ka , V, F, and X 0 of the two models are set to the same values), the two absorption curves almost overlap. Even if the distribution phase in the two-compartment model causes the drug time curve to decrease faster, it will not have an impact on the absorption fraction. Therefore, the key to analyzing the absorption kinetics of drugs should be the absorption phase before T max .
  • the increase in blood drug concentration is caused by first-order rate absorption, while the concentration decrease is caused by first-order rate elimination; for the two-compartment model, the increase in blood drug concentration is caused by first-order rate absorption, The decrease in blood drug concentration is caused by a first-order rate processing process (the distribution and elimination rates are both first-order rate processes).
  • the drug-time curve can be simplified into two parts: a linear kinetic increase process and a linear kinetic decrease process of blood drug concentration over time. At this time, the relationship expression between blood drug concentration C and time t should be:
  • k a is the linear kinetic rising rate constant (i.e., the absorption rate constant of the drug); k is the linear kinetic falling rate constant; A is the correction coefficient; Formula 1 is not dependent on the compartment model, but is only related to the drug time curve characteristics The Ct relationship of the relevant in vivo absorption kinetic model.
  • An iterative method (the code is written in Python 3.6.7 software) is used to analyze the k a value of the absorption kinetic model that does not depend on the compartment model.
  • the k and k a values are continuously traversed and brought into Formula 1 according to the set value range (for example, k a is 0.01 ⁇ 10.0h -1 , k is 0.01 ⁇ 5.0h -1 ), and the value is calculated according to the value of every 0.01 Iterate and obtain several drug time curves. Since the calculation of the A value is related to the number of points taken after reaching the peak, the blood drug concentration data after reaching the peak should be gradually increased until the sum of the absolute values of the residual values of the estimated blood drug concentration data and the measured blood drug concentration data is minimized (Formula 3), Output the best k a value;
  • C i is the measured blood drug concentration
  • C i' is the estimated blood drug concentration.
  • This k a value is the k a estimate of the absorption kinetics model in which the in vivo absorption kinetics does not depend on the compartment model.
  • the running time of each set of blood drug concentration data is approximately within 2 minutes.
  • Reagents isopropyl alcohol, acetic acid, ammonium acetate, acetonitrile.
  • the above reagents are all chromatography grade.
  • Test drug carbamazepine tablets-reference preparation ( Specification: 100mg), Sun Pharmaceutical Industries Ltd; Carbamazepine tablets - test preparation (Specification: 100mg), provided by a domestic pharmaceutical company; Cyclosporine soft capsule - reference preparation (Sandimmun Specification: 50mg), Novartis Pharma Sau AG; Cyclosporine soft capsule-test preparation (specification: 50mg), provided by a domestic pharmaceutical company.
  • the accuracy of the absorption kinetic model and its analysis method that does not depend on the compartment model is verified through the pharmacokinetic parameter setting of the compartment model and clinical trial data.
  • the single-compartment model parameters and the dual-compartment model parameters were randomly set to obtain the corresponding characteristic drug-time curve.
  • the blood drug concentration data that satisfies the single-compartment model uses the Wagner-Nelson method and the in vivo absorption kinetic model that is independent of the compartment model to calculate the k a value
  • the dual-compartment model uses the Loo-Riegelman method, which is independent of the compartment model. Calculate the k a value using the model's absorption kinetic model method and the statistical method of moments.
  • carbamazepine tablets in line with the single-compartment model
  • cyclosporine soft capsules in line with the dual-compartment model
  • Analyze the estimated k a value of the corresponding drug preparation, and conduct correlation analysis with pharmacokinetic parameters such as C max and T max that reflect the absorption characteristics in the body, to further verify the absorption kinetic model and its analysis method that do not rely on the compartment model. Accuracy and practical application value.
  • the k value is smaller than the ka value.
  • the effects of changes in the single-compartment model k a and k value on the accuracy of the absorption kinetic model analysis k a that does not depend on the compartment model were separately investigated, divided into two situations: 1 Keep the k value unchanged (set to 0.10h -1 ), randomly change the ka value (range is 0.15 ⁇ 5.00h -1 , take value every 0.05h -1 ); 2Keep the ka value unchanged (set to 3.00h -1 ), randomly change the k value (The range is 0.01 ⁇ 2.01h -1 , and the value is taken every 0.05h -1 ). Calculate the blood drug concentrations at different time points according to Formula 5, and obtain multiple sets of characteristic drug-time curves that satisfy the single-compartment model.
  • ⁇ represents the first-order rate constant of distributed phase mixing
  • ⁇ represents the first-order rate constant of elimination phase mixing, which are calculated by Formula 7 and Formula 8 respectively;
  • Carbamazepine tablets and cyclosporine soft capsules were used as model drugs to conduct human pharmacokinetic tests. The study was approved by the Medical Ethics Committee of Xiangya School of Pharmacy, Central South University.
  • venous blood samples were collected at 0h before administration and at 1h, 2h, 3h, 4h, 5h, 6h, 7h, 8h, 9h, 10h, 12h, 14h, 24h, 36h, 48h, and 72h after administration. in vacuum blood collection tubes containing heparin sodium anticoagulant. Blood samples were centrifuged at 1700 g for 10 min at 4°C to separate plasma. Plasma samples were stored in a -70°C ultra-low temperature refrigerator.
  • Mass spectrometry conditions electrospray ion source (ESI), positive ion multiple reaction monitoring mode, the detection ion transition of carbamazepine is 237.1 ⁇ 194.2 (m/z), and the detection ion transition of the internal standard carbamazepine-d 8 is 245.2 ⁇ 202.1(m/z).
  • Each cycle is performed at 0h before dosing and 0.5h, 0.75h, 1h, 1.25h, 1.5h, 1.75h, 2h, 2.25h, 2.5h, 3h, 4h, 6h, 8h, 10h, 12h, and 14h after dosing, respectively.
  • Whole blood samples were stored in a -70°C refrigerator.
  • the obtained blood concentration data of carbamazepine tablets and cyclosporine soft capsules were calculated using WinNonlin 8.2 traditional pharmacokinetic mode to determine the compartmental models of carbamazepine and cyclosporine; and then the NCA mode was used to calculate the AIC values.
  • Pharmacokinetic parameters such as C max , T max , AUC 0-t , AUC 0- ⁇ and elimination half-life (t 1/2 ) of the model drug.
  • the blood drug concentration data obtained after setting parameter values for the single-compartment model and the dual-compartment model, and the clinical pharmacokinetic data of carbamazepine tablets and cyclosporine soft capsules were input into the Python iterative method program (Appendix A).
  • the value range of ka is 0.01 ⁇ 10.0h -1
  • the value range of k is 0.01 ⁇ 5.0h -1 . You can get the value of ka by running the program.
  • the Loo-Riegelman method is used to calculate the ka value of the dual-compartment model parameter setting group and the clinical pharmacokinetic data (cyclosporine soft capsules) that satisfies the dual-compartment model, as an absorption kinetics model that is independent of the compartment model. comparative study.
  • (X p ) t /V c represents the amount of drug entering the peripheral chamber at time t.
  • ⁇ c and ⁇ t represent the blood drug concentration difference and time interval between two consecutive samples, respectively. Therefore, perform linear regression on F abs and t to obtain the straight line equation, and its slope is the k a value (Formula 13).
  • C i , C i+1 and C n represent the drug concentration at time points t i , t i+1 and t n respectively;
  • MAT is the average absorption time;
  • MRT is the average residence time of the drug in the body;
  • k T is the terminal elimination Rate constant;
  • AUMC represents the area under the curve of time-plasma concentration product and time.
  • the value of k a is 0.15 ⁇ 5.00h -1 (that is, the absorption half-life t 1/2, abs is 0.14 ⁇ 4.62h), and the value of k is 0.01 ⁇ 2.01h -1 (that is, the elimination half-life t 1/2 is 0.34 ⁇ 69.30 h), satisfying the ka and k value ranges of most single-compartment model drugs. Under different parameter values, the accuracy of calculating the ka value was compared between the absorption kinetic model method that does not depend on the compartment model and the Wagner-Nelson method.
  • the k setting value is changed from 0.01
  • the k a value calculated using the absorption kinetic model that does not depend on the compartment model is basically consistent with the true value of k a in the drug-time curve.
  • the accuracy of estimating the k a value by the Wagner-Nelson method increases with the The k setting value gradually decreases as the k value increases.
  • the accuracy of the Wagner-Nelson method is lower than 85%.
  • the absorption kinetic model that does not rely on the compartment model should have a higher accuracy in estimating the k a value of the single-compartment model drug, because when the V, F, and X 0 of the single-compartment model are set to fixed values, the blood
  • the relationship between drug concentration C and time t is basically consistent with the Ct relationship (Formula 1) of the absorption kinetics model that does not depend on the compartment model.
  • the human blood drug concentration data of 36 different drug preparations were obtained from the literature, and WinNonlin software was used to calculate the AIC value of each drug.
  • the results are shown in Table 1.
  • the AIC 2 values (dual-compartment model) of all drugs are smaller than the AIC 1 values (single-compartment model), indicating that the in vivo processes of the 36 drugs are consistent with the dual-compartment model.
  • the WinNonlin software was used to preliminarily estimate the k a (0.210 ⁇ 1.826h -1 ), k 12 (0.044 ⁇ 0.847h -1 ), k 21 (0.010 ⁇ 0.451h -1 ) and k 10 (0.012 ⁇ 1.003h) of each drug.
  • -1 Scope The human blood drug concentration data of 36 different drug preparations were obtained from the literature, and WinNonlin software was used to calculate the AIC value of each drug. The results are shown in Table 1.
  • the AIC 2 values (dual-compartment model) of all drugs
  • the sum of the k 12 and k 10 values of all drugs is less than the k a value (i.e., k a > k 12 + k 10 ), and the k a value of each drug is greater than the k 12 value, and
  • the k 12 values are all higher than the k 21 values (that is, k a > k 12 > k 21 ).
  • the k 10 values are significantly higher than k 21 (p ⁇ 0.05). This result satisfies the k a and k a of the two-compartment model drugs.
  • the parameter settings of k 10 , k 12 and k 21 provide important basis.
  • Table 1 Human blood drug concentration data and analytical pharmacokinetic parameters of 36 pharmaceutical preparations obtained from the literature
  • the results are shown in Table 2.
  • the k a value estimated using the absorption kinetic model independent of the compartment model has positive and negative RE values, and all RE values are within ⁇ 16%. Most of the RE values are within ⁇ 10%, indicating that the absorption kinetics model that does not depend on the compartment model has a high accuracy in analyzing the ka value of the dual-compartment model drug.
  • the k a value estimated by the Loo-Riegelman method has a larger change in the RE value, and all of them are positive values (i.e.
  • NA MAT is a negative value and cannot be calculated.
  • the k a value estimated using an absorption kinetic model that is independent of the compartment model is not affected by changes in parameters such as k 12 , k 21 and k 10 , and maintains good accuracy, indicating that it is not dependent on the compartment model.
  • the absorption kinetic model does not depend on the compartment model parameters; since the Loo-Riegelman method is a classic method of the two-compartment model, parameters such as k 12 , k 21 and k 10 are used in the calculation process of this method, so the values of these parameters change It is more sensitive to the accuracy of k a , but still has better accuracy than the statistical moment method; the statistical moment method is a non-compartmental model method and is almost not affected by changes in k 12 , k 21 and k 10 , but its accuracy is lower Difference.
  • Carbamazepine and cyclosporine are both drugs with narrow therapeutic windows, so the clinical BE trial is a two-sequence, four-cycle self-crossover trial design.
  • the human body drug time curves after oral administration of carbamazepine tablets and cyclosporine soft capsules in fasting and postprandial states are shown in Figure 3, and the pharmacokinetic parameters are summarized in Table 3.
  • the drug-time curves of the reference preparation and the test preparation under fasting and postprandial conditions were relatively close. After fasting administration, the T max of the two preparations was approximately 3.0 hours, while the peak time after meals was delayed to 4.9 hours.
  • the drug-time curves of the reference preparation and the test preparation under fasting and postprandial conditions were relatively close. After fasting administration, the T max of the two preparations was approximately 1.3 hours, while the peak time after meals was delayed to 2.5 hours.
  • Table 4 Calculate the k a value of carbamazepine tablets using the absorption kinetic model method and the Wagner-Nelson method that do not depend on the compartment model
  • the absorption kinetic model method that does not depend on the compartment model and the Loo-Riegelman method were used to calculate the fasting time of the reference preparation and the test preparation of cyclosporine soft capsules. and the ka value in the postprandial state.
  • the k 10 , k 12 , and k 21 parameter values required by the Loo-Riegelman method are obtained from the venous blood drug concentration analysis in the literature; in addition, due to the poor accuracy of the statistical moment method, it has not been applied to cyclosporine soft tissue.
  • Capsule ka analysis The ka analysis results of different methods are shown in Table 5.
  • the ka values of the cyclosporine reference preparation and the test preparation estimated by the absorption kinetic model method that does not depend on the compartment model and the Loo-Riegelman method are in the same state. There was no significant difference under the condition. Due to the influence of food, the ka values of the two preparations in the postprandial state were significantly lower than those in the fasting state (p ⁇ 0.001).
  • the correlation coefficient R>0.93 between the k a value estimated using the Loo-Riegelman method and T max , C max , C max /AUC 0-t is R>0.93, but k a
  • the p-values with the three pharmacokinetic parameters are slightly higher than 0.05, indicating potential correlation.
  • the drug-time curve is fitted by iteratively taking values of k a and k until the sum of the absolute values of the residuals of the measured drug-time curve is Minimum. If the absorption phase and C max of the fitted drug-time curve are closer to the actual measured blood drug concentration data, it means that the accuracy of the k a estimation is higher.
  • the present invention verifies the accuracy of the absorption kinetic model and its analysis method that do not depend on the compartment model through single-compartment and dual-compartment model parameter settings and clinical measured data respectively.
  • the verification results of the compartment model parameters show that the accuracy of the analytical k a value of the absorption kinetic model that does not depend on the compartment model is slightly better than Wagner-Nelson, better than Loo-Riegelman and the statistical moment method, and the calculation process and its accuracy It has nothing to do with the model parameters of each compartment.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Chemical & Material Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Medicinal Chemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Mathematics (AREA)
  • Pharmacology & Pharmacy (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Medical Informatics (AREA)
  • Operations Research (AREA)
  • Epidemiology (AREA)
  • Algebra (AREA)
  • Toxicology (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

本发明公开了一种药物吸收速率常数的预测模型、设备和存储介质，通过构建不依赖于房室模型的吸收动力学模型解析得到估算ka值，依据现有房室模型获得真实ka值，通过估算ka值和真实ka值验证比对差异，基于药动学参数进一步验证ka值的准确性，本发明预测模型预测的ka值准确度高，且无需静脉血药浓度数据，可满足不同类型药物的药物吸收速率常数ka值估算，进而为药物制剂的体内吸收动力学解析及其IVIVC的体内评价提供重要的工具。

Description

药物吸收速率常数的预测模型、设备和存储介质 技术领域
本发明涉及药物信息技术领域，具体涉及一种药物吸收速率常数的预测模型、设备和存储介质。
背景技术
药物吸收是指药物自给药部位摄取进入血液循环的过程。药物的吸收速率，是药代动力学和生物利用度研究的一个重要参数。吸收速率对血药浓度影响很大，它受给药途径和许多其他因素的影响。给药途径不同，吸收速率自然不同；剂型不同，吸收速率亦各异。如固体剂型的口服吸收，就取决于制剂的崩解、活性药物的溶出、在吸收部位的药物浓度和血液循环，以及吸收面的位置和面积。若药物不吸收，当然不能起全身作用；如果吸收差，则必须加大剂量；要是吸收慢，可能作用会延迟；吸收过快，可发生副作用；吸收不规则，又可能无法预测临床效应。因此，了解药物吸收速率，是临床前和临床研究的一个主要组成部分。
药物吸收速率常数ka值是表示药物在吸收部位进入血液系统速率的一个相对指标，也是有吸收药物动力学房屋模型中的一个主要模型参数。在计算药物在体内达到的峰值时间tmax和药物在体内的峰值浓度Cmax值中，ka值也起着主导作用，并且在给药剂量方案的确定，缓冲长效剂的缓释速率计算中也有比较重要的意义。估算药物吸收速率常数Ka的常用定量方法有房室模型法(Wagner-Nelson法、Loo-Riegelman法)和非房室模型法(反卷积分法和统计矩法)。对于给定药物制剂的药时曲线，Wagner-Nelson法和Loo-Riegelman法估算ka的准确性显著高于非房室模型法，但需准确解析房室模型参数，同时多种药物静脉数据缺乏导致Loo-Riegelman无法应用等。
基于此，有必要提供一种无需静脉血药浓度数据的前提下，满足不同类型药物的药物吸收速率常数ka值估算和准确性的验证方法。
发明内容
针对现有技术的不足，本发明的目的在于提供一种药物吸收速率常数的预测模型、设备和存储介质，通过构建不依赖于房室模型的吸收动力学模型解析得到估算ka值，依据现有房室模型的获得真实ka值，通过估算ka值和真实ka值验证比对差异，基于药动学参数进一步验证ka值的准确性，本发明预测模型预测的ka值准确度高，且无需静脉血药浓度数据，可满足不同类型药物的药物吸收速率常数ka值估算，进而为药物制剂的体内吸收动力学解析及其IVIVC的体内评价提供重要的工具。
为实现上述目的，本发明所采用的技术方案具体如下：
一方面，本发明提供一种药物吸收速率常数的预测模型，所述模型包括：
获取模块，用于获取药物制剂的实测血药浓度、时间采样点，进而绘制实测药时曲线，获得药物体内吸收特性的药动学参数(如Cmax、Tmax、AUC0-t、AUC0-∞和消除半衰期(t1/2))；
拟合模块，根据实测药时曲线、Tmax，输出不依赖于房室模型的吸收动力学模型，并获得估算ka值；
真实模块，根据血药浓度特征选择房室模型预输入函数、实测血药浓度、时间采样点、固定参数，得到真实ka值；
比对模块，对比估算ka值与真实ka值的差异，输出准确性高的估算ka值。
进一步地，还包括修正模块，利用估算ka值预测药物制剂的吸收相与预测Cmax值，比对真实Cmax和预测Cmax，输出最佳估算ka值。
进一步地，所述不依赖于房室模型的吸收动力学模型的建立过程如下：
基于实测药时曲线，以Tmax对应的血药浓度点，将实测药时曲线分为血药浓度随时间的线性动力学上升和线性动力学下降两个过程，实测药时曲线的吸收相进行反卷积分获取药物体内吸收曲线，构建血药浓度与时间C-t的表达关系式，从而建立不依赖于房室模型的吸收动力学模型，其中，C-t表达关系式为：
其中，ka为线性动力学上升速率常数；k为线性动力学下降速率常数；A为校正系数。
本发明构建的公式1为不依赖于房室模型、仅与药时曲线特征有关的体内吸收动力学模型的C-t关系式。
进一步地，所述估算ka值解析过程如下：
步骤1：基于药物制剂的实测血药浓度，血药浓度C经对数化后，取达峰后的连续若干数值，拟合线性回归，线性方程为LnC＝-kt+LnA(公式2)，其中截距为LnA，获得A值；
步骤2：基于公式1和步骤1获得的A值，运用Python迭代法程序将k值与ka值按照设定的数值范围(如ka为0.01～10.0h-1，k为0.01～5.0h-1)持续遍历带入公式1，并按照每0.01的数值进行迭代，以得到与实测药时曲线残差绝对值之和最小的预测药时曲线；
步骤3：运用公式3计算步骤2获得的药时曲线中药物制剂的预估血药浓度与实测血药浓度的残差绝对值之和最小，进而输出最佳ka值；
其中，Ci为实测的血药浓度，Ci’为估算的血药浓度。SUM值越小，表明拟合度越好。
进一步地，所述最佳ka值为体内吸收不依赖于房室模型的吸收动力学模型的估算ka值。
进一步地，所述真实ka值的获取过程如下：
方式1：
基于单室模型血管外给药的血药浓度计算公式5，
其中，ka为线性动力学上升速率常数，即药物吸收速率常数；k为线性动力学下降速率常数，即消除速率常数。
将X0、F、V随机设置为固定值(如X0＝5.0mg，F＝1，V＝50L)，分为两种情况：①保持k值不变(设定为0.10h-1)，随机变化ka值(范围为0.15～5.00h-1，每隔0.05h-1取值)；②保持ka值不变(设定为3.00h-1)，随机变化k值(范围为0.01～2.01h-1，每隔0.05h-1取值)；计算不同时间点的血药浓度，获得满足单室模型的多组特征药时曲线，获得真实ka值；
方式2：
基于双室模型血管外给药的血药浓度计算公式6～公式8，将X0、F、Vc随机设置为固定值(如X0＝2.2mg，F＝1，Vc＝10L)；从现有技术中获取满足双室模型药物的血药浓度数据；使用WinNonlin软件(8.2版本，Certara公司)初步计算这些药物的ka、k12、k21和k10值，并比较各参数间的关系；
将所获得的ka、k12、k21和k10值按降序排序，再将这些参数的前1/3、中1/3和后1/3(n＝12)的平均值分别设置为高、中、低数值水平；每个参数的不同级别间随机组合，再将这些参数(ka、k12、k21和k10)带入公式6～公式8，进而获得不同时间点(间隔0.1h)的血药浓度，绘制药时曲线；
其中，ka为线性动力学上升速率常数，即药物吸收速率常数；k为线性动力学下降速率常数，即消除速率常数；k12是指双室模型中药物从中央室(血液)向周边室(器官、组织)扩散的速率常数，k21是指双室模型中药物从周边室(器官、组织)向中央室(血液)扩散的速率常数，k10是指双室模型中药物从中央室(血液)消除的速率常数。
公式6中，α表示分布相混合一级速率常数，β表示消除相混合一级速率常数，分别由公式7、公式8计算；

获得真实ka值。
进一步地，方式1中，对单室模型来说，当k＝0.10h-1不变，ka设定值由0.15～5.00h-1变化时，药时曲线采用本发明所述模型计算所得ka值与真实ka值基本一致；当ka＝3.00h-1不变，k设定值由0.01～2.01h-1变化时，药时曲线采用不依赖于房室模型的吸收动力学模型计算所得ka值与真实ka值基本一致。
结果表明，估算ka值的准确性与单室模型参数取值无关(即不依赖于房室模型参数)，仅与药时曲线的形状有关。
进一步地，所述满足双室模型药物为醋酸阿比特龙片、阿昔洛韦混悬剂、阿奇霉素片、苯那普利胶囊、安非他酮片、坎地沙坦酯片、卡托普利片、塞来昔布胶囊、环丙沙星片、氯吡格雷片、达卡他韦片、多潘立酮片、屈他维林片、格列本脲片、氢氯噻嗪片、伊拉地平胶囊、伊曲康唑片、拉西地平片、盐酸乐卡地平片、左炔诺孕酮片、氯雷他定片、二甲双胍片、吗替麦考酚片、萘普生片、奥美沙坦酯片、磷酸奥司他韦胶囊、喹那普利片、瑞格列奈片、利匹韦林片、瑞舒伐他汀片、西洛多辛胶囊、辛伐他汀片、替米沙坦片、富马酸替诺福韦酯片、特比萘芬片、替格瑞洛片。
进一步地，方式2中，对双室模型来说，基于满足双室模型药物的ka、k12、k21、k10值，按照降序排列，分别获取高、中、低三个水平的参数值，其中ka为1.098、0.603、0.375h-1；k12为0.525、0.211、0.133h-1；k21为0.176、0.067、0.025h-1；k10为0.571、0.271和0.100h-1；根据ka、k12、k21、k10之间的关系(即ka＞k12+k10，且ka＞k12＞k21)，随机组合ka、k12、k21和k10的数值，并根据公式6～公式8计算每组的药时曲线；采用不依赖于房室模型的吸收动力学模型的估算ka值与真实ka值相比，RE有正值和负值，所有RE值均在±16％以内，其中大部分RE值在±10％以内，表明不依赖于房室模型的吸收动力学模型解析的ka值的准确度较高。
结果证明了不依赖于房室模型的吸收动力学模型的参数(k12、k21、k10等)，无需静脉血药浓度数据，该不依赖于房室模型的吸收动力学模型估算的ka值具有较高的准确性，满足不同类型药物的ka值解析。进一步地，所述药物制剂为卡马西平片或环孢素软胶囊。
一方面，本发明提供一种药物吸收速率常数准确性的验证方法，通过对药物制剂的血药浓度分析绘制实测药时曲线；基于实测药时曲线建立不依赖于房室模型的吸收动力学模型，计算相应模型的吸收速率常数ka；基于现有房室模型参数设定和临床试验数据等验证不依赖于房室模型的吸收动力学模型计算的吸收速率常数ka的准确性。
进一步地，所述验证方法，包括以下步骤：
S1.采集患者给予药物制剂后的血浆样本，经HPLC方法测定药物制剂的血药浓度；基于血药浓度和采 样时间点，获得药物动力学参数(例如Cmax、Tmax、AUC0-t、AUC0-∞和消除半衰期(t1/2))，绘制实测药时曲线；
S2.基于实测药时曲线，以Tmax对应的血药浓度点，建立不依赖于房室模型的吸收动力学模型，
其中，ka为线性动力学上升速率常数(即药物的吸收速率常数)；k为线性动力学下降速率常数；A为校正系数；
S3.基于采用迭代法解析不依赖于房室模型的吸收动力学模型的估算ka值；
S4.根据血药浓度特征选择房室模型并设定得ka值，所得ka值为真实ka值；
S5.对比估算ka值与真实ka值的差异，准确性以RE表示，计算公式为：RE％＝(估算ka值-真实ka值)/真实ka值×100％(公式4)。RE越小，估算ka值越准确。
进一步地，还包括S6.将药动学参数(Cmax、Tmax、Cmax/AUC0-t等反映体内吸收特征的药动学参数)与估算ka值进行Pearson相关性分析(SPSS 25.0，SPSS Inc.)，进一步验证ka值的准确性。
进一步地，通过运用估算ka值计算估算Cmax，估算Cmax与实测Cmax越接近，代表估算ka值的准确度越高。估算ka值与反映药物体内吸收特性的药动学参数(Cmax、Tmax等)均具有良好的相关性，还可准确预测药物的吸收相与Cmax值。
进一步地，基于S1步骤属于常规方法，不是本申请的保护重点，此处不再一一赘述了。
另一方面，本发明提供一种药物吸收速率常数的获取方法，包括以下步骤：
1).采集患者给予药物制剂后的血浆样本，经HPLC方法测定药物制剂的血药浓度；基于血药浓度和采样时间点，获得药物动力学参数(例如Cmax、Tmax、AUC0-t、AUC0-∞和消除半衰期(t1/2))，绘制实测药时曲线；
2).基于实测药时曲线，以Tmax对应的血药浓度点，建立不依赖于房室模型的吸收动力学模型，
其中，ka为线性动力学上升速率常数(即药物的吸收速率常数)；k为线性动力学下降速率常数；A为校正系数；
3).基于采用迭代法解析不依赖于房室模型的吸收动力学模型的估算ka值。
一方面，本发明提供一种运用本发明建立的不依赖于房室模型的吸收动力学模型预测药物制剂的Cmax的方法，步骤如下：
运用本发明所述的验证方法获得估算ka值，通过ka与k的迭代取值以拟合药时曲线，直至与实测药时曲线的残差绝对值之和最小，获得预测Cmax值。
若拟合的药时曲线吸收相和预测Cmax值与实测血药浓度数据越接近，代表ka估算的准确性越高。
另一方面，本发明提供一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序是实现本发明所述验证方法的步骤。
另一方面，本发明提供一种计算机可读存储介质，其存储有计算机程序，所述计算机程序被处理器执行实现本发明所述验证方法的步骤。
与现有技术相比较，本发明的有益技术效果如下：
1：本发明综合考虑了药物的血药浓度、采样时间点、单剂量、总药量、Cmax、Tmax、AUC0-t、AUC0-∞、消除半衰期(t1/2)等建模关键药动学参数，构建了不依赖于房室模型的吸收动力学模型。将该吸收动力学模型、Wagner-Nelson法、Loo-Riegelman法、反卷积分法和统计矩法估算得到的估算ka值和房室模型设定得到的真实ka值，经比对各种方法验证估算ka值的准确性，实现准确反映药物在体内的吸收相和Cmax。
2：本发明方法构建的不依赖与房室模型的吸收动力学模型解决传统的方法估算ka值的弊端，创造性地提出了不依赖房室模型、仅与药时曲线特征有关的体内吸收动力学模型的公式1，提高解析估算ka值的准确度。
3：本发明提供的药物制剂药时曲线预测模型解析的ka值准确性高，适用范围广泛，且能为药物制剂的体内吸收动力学解析及其体内外相关性(IVIVC)的体内评价提供重要的工具。
4：本发明方法成功应用于卡马西平片和环孢素软胶囊两个模型药物的ka解析，且与反映药物体内吸收特性的药动学参数(Tmax、Cmax、Cmax/AUC0-t等)均具有良好的相关性，还可准确预测药物的吸收相与Cmax，还可用于参比制剂和受试制剂的质量控制。
附图说明
图1：不依赖于房室模型的吸收动力学模型与Wagner-Nelson法估算单室模型药物ka值的准确性。(A)k不变，ka值变化(0.15～5.00h-1)；(B)ka不变，k值变化(0.01～2.01h-1)。
图2：双室模型参数设置组的药时曲线(共39组数据)。
图3：卡马西平片和环孢素软胶囊口服给药后的人体内药时曲线。(A)卡马西平片参比制剂与受试制剂(平均值±标准差，n＝60)；(B)环孢素软胶囊参比制剂与受试制剂(平均值±标准差，空腹n＝46，餐后n＝90)
图4药物制剂实测与预测的平均药时曲线。卡马西平片的(A)空腹-参比制剂，(B)空腹-受试制剂，(C)餐后-参比制剂，(D)餐后-受试制剂；环孢素软胶囊的(E)空腹-参比制剂，(F)空腹-受试制剂，(G)餐后-参比制剂，(H)餐后-受试制剂。
具体实施方式
下面对本发明的技术方案进行举例说明，本发明请求保护的范围包括但不限于以下实施例。
实施例
1药物制剂体内吸收动力学不依赖于房室模型的吸收动力学模型的建立
药物以单剂量X0给药后，到达吸收部位的总药量为Xa。在单室模型中，药物以一级速率过程吸收(ka)，进入人体药量为X，再以一级速率过程进行消除(k)；在双室模型中，药物吸收进入中央室(血液)后以一级速率过程向周边室(器官、组织)分布(k12)与消除(k10)，周边室药物也以一级速率过程返回至中央室(k21)。无论单室模型和双室模型药物，药时曲线均以Tmax为界，可划分为吸收相和处置相(其中单室模型等同消除相；双室模型为分布相与消除相之和)。在吸收相中，药物的吸收速率始终大于处置速率，导致血药浓度持续上升；达到Tmax时，药物吸收速率与处置速率相等；此后将以分布和/或消除为主，使得血药浓度持续下降直至消除结束。将单室模型和双室模型药物的特征药时曲线(两个模型的ka、V、F、X0设置为相同数值)的吸收相反卷积分后，两吸收曲线几乎重叠。即使双室模型中存在分布相而导致药时曲线下降变快，但对于吸收分数也不会产生影响。因此，药物的吸收动力学解析的关键应为Tmax之前的吸收相。
对于单室模型而言，血药浓度升高过程为一级速率吸收导致，而浓度下降过程为一级速率消除导致；对于双室模型而言，血药浓度升高为一级速率吸收导致，而血药浓度下降为一级速率处置过程(分布和消除速率均为一级速率过程)导致。当不考虑房室模型，药时曲线可以简化为血药浓度随时间的线性动力学上升过程和线性动力学下降过程两个部分。此时，血药浓度C与时间t的关系表达式应为：
其中，ka为线性动力学上升速率常数(即药物的吸收速率常数)；k为线性动力学下降速率常数；A为校正系数；公式1为不依赖于房室模型、仅与药时曲线特征有关的体内吸收动力学模型的C-t关系式。
2药物制剂体内吸收速率常数的解析方法建立
采用迭代法(代码写入Python 3.6.7软件)解析不依赖于房室模型的吸收动力学模型的ka值。解析步骤如下：采用给定药物的达峰前全部血药浓度数据和达峰后部分血药浓度数据，将血药浓度C对数化(即LnC)，取达峰后的连续若干数值，作线性回归，线性方程为LnC＝-kt+LnA(公式2)，其中截距为LnA，获得A值。同时，k与ka值按照设定的数值范围(如ka为0.01～10.0h-1，k为0.01～5.0h-1)持续遍历带入至公式1中，并按照每0.01的数值进行迭代，获得若干条药时曲线。由于A值计算与达峰后取点数有关，应逐步增加达峰后血药浓度数据，直至达到估算的血药浓度数据与实测血药浓度数据的残差绝对值之和最小(公式3)，输出最佳ka取值；
其中，Ci为实测的血药浓度，Ci’为估算的血药浓度。SUM值越小，表明拟合度越好。该ka值即为体内吸收动力学不依赖于房室模型的吸收动力学模型的ka估算值。每组血药浓度数据的运行时间约为2min以内。
3不依赖于房室模型的吸收动力学模型及其解析方法的验证
3.1仪器与材料
3.1.1仪器：6470型三重四极杆液质联用仪、TDL5型台式低温冷冻离心机、DW-86L828J型超低温冰箱
3.1.2试剂：异丙醇、醋酸、醋酸铵、乙腈，以上试剂均为色谱级。
3.1.3试药：卡马西平片-参比制剂(规格：100mg)，Sun Pharmaceutical Industries Ltd；卡马西平片-受试制剂(规格：100mg)，由国内某药企提供；环孢素软胶囊-参比制剂(Sandimmun规格：50mg)，Novartis Pharma Schweiz AG；环孢素软胶囊-受试制剂(规格：50mg)，由国内某药企提供。
3.2实验方法
通过房室模型药动学参数设定和临床试验数据验证不依赖于房室模型的吸收动力学模型及其解析方法的准确性。具体地，随机设定单室模型参数与双室模型参数，获取相应的特征药时曲线。其中，满足单室模型的血药浓度数据采用Wagner-Nelson法和体内不依赖于房室模型的吸收动力学模型计算ka值，而双室模型则采用Loo-Riegelman法、不依赖于房室模型的吸收动力学模型的方法及统计矩法计算ka值，对比不同方法估算的ka值与真实ka值(即设定ka值)的差异，明确不依赖于房室模型的吸收动力学模型及方法解析ka的准确度。此外，对比各房室模型参数变动对不依赖于房室模型的吸收动力学模型解析的ka准确性影响，明确不依赖于房室模型的吸收动力学模型与房室模型参数之间的关联。另一方面，卡马西平片(符合单室模型)和环孢素软胶囊(符合双室模型)，分别通过不依赖于房室模型的吸收动力学模型、Wagner-Nelson法或Loo-Riegelman法解析相应药物制剂的估算ka值，并与Cmax和Tmax等反映体内吸收特征的药动学参数进行相关性分析，进一步验证不依赖于房室模型的吸收动力学模型及其解析方法的准确性和实际应用价值。
3.2.1房室模型药动学参数的设定
3.2.1.1单室模型参数的设定
基于单室模型血管外给药的血药浓度计算公式5，
将X0、F、V随机设置为固定值(如X0＝5.0mg，F＝1，V＝50L)。通常情况下，k值小于ka值。分别考察单室模型ka与k值的变化对不依赖于房室模型的吸收动力学模型解析ka准确性的影响，分为两种情况：①保持k值不变(设定为0.10h-1)，随机变化ka值(范围为0.15～5.00h-1，每隔0.05h-1取值)；②保持ka值不变(设定为3.00h-1)，随机变化k值(范围为0.01～2.01h-1，每隔0.05h-1取值)。分别按照公式5计算不同时间点的血药浓度，获得满足单室模型的多组特征药时曲线。
3.2.1.2双室模型参数的设定
基于双室模型血管外给药的血药浓度计算公式6～公式8，将X0、F、Vc随机设置为固定值(如X0＝2.2mg，F＝1，Vc＝10L)。由于ka、k12、k21和k10取值范围大，且数据组较多。为了排除一些无效参数设置组，本研究通过GetData Graph Digitizer软件(2.25版本)从文献中获取36个满足双室模型药物的血药浓度数据。使用WinNonlin软件(8.2版本，Certara公司)初步计算这些药物的ka、k12、k21和k10值，并比较各参数间的关系。
将所获得的36个药物的ka、k12、k21和k10值按降序排序，再将这些参数的前1/3、中1/3和后1/3(n＝12)的平均值分别设置为高、中、低数值水平。每个参数的不同级别间随机组合，再将这些参数(ka、k12、k21和k10)带入公式6～公式8，进而获得不同时间点(间隔0.1h)的血药浓度，绘制药时曲线。
公式6中，α表示分布相混合一级速率常数，β表示消除相混合一级速率常数，分别由公式7、公式8计算；

获得真实ka值。
3.2.2卡马西平片和环孢素软胶囊的临床数据获取
以卡马西平片和环孢素软胶囊为模型药物进行人体药动学试验，该研究获得了中南大学湘雅药学院医学伦理委员会批准。
(1)卡马西平片空腹与餐后BE试验设计及血药浓度检测方法
本研究采用单中心、随机、开放、两序列、四周期自身交叉试验设计，进行空腹和餐后BE研究。空腹和餐后临床试验均入组30例健康受试者，并签署书面知情同意书。每周期空腹/餐后(高脂餐含142kcal蛋白质、256kcal碳水化合物、569kcal脂肪，总热量约967kcal)单次口服1片卡马西平片受试制剂(规格：100mg)或1片卡马西平片参比制剂(规格：100mg)，温水240mL送服。周期间的清洗 期为21天。每周期于给药前0h和给药后1h、2h、3h、4h、5h、6h、7h、8h、9h、10h、12h、14h、24h、36h、48h、72h分别采集静脉血样约4mL，置于含肝素钠抗凝剂的真空采血管中。血样于4℃下，以1700g离心10min，分离血浆。将血浆样本置于-70℃超低温冰箱中保存。
取40μL血浆样品，加入10μL卡马西平-d8工作溶液(内标，2.0μg/mL)混匀，加入纯乙腈150μL，涡旋2min，以15000rpm离心5min，取上清液5μL注入高效液相串联质谱仪(High-performance liquid chromatography tandem mass spectrometry，HPLC-MS/MS)系统进行分析。色谱条件为C18柱(2.1mm×50mm，3.5μm，Waters)；流动相为0.2％醋酸水溶液-乙腈＝2∶1(v/v)，等度洗脱，流速0.5mL/min；柱温30℃。质谱条件：电喷雾离子源(ESI)，正离子多重反应监测模式，卡马西平的检测离子对为237.1→194.2(m/z)，内标卡马西平-d8的检测离子对为245.2→202.1(m/z)。
(2)环孢素软胶囊空腹与餐后BE试验设计及血药浓度检测方法
本研究采用单中心、随机、开放、两序列、四周期自身交叉试验设计，进行空腹和餐后BE研究。空腹临床试验入组23例健康受试者，餐后临床试验入组45例健康受试者，并签署书面知情同意书。每周期空腹/餐后(高脂餐含149kcal蛋白质、288kcal碳水化合物、521kcal脂肪，总热量约958kcal)单次口服1粒环孢素软胶囊受试制剂(规格：50mg)或1粒参比制剂(Sandimmun规格：50mg)，温水240mL送服。周期间的清洗期为7天。每周期于给药前0h和给药后0.5h、0.75h、1h、1.25h、1.5h、1.75h、2h、2.25h、2.5h、3h、4h、6h、8h、10h、12h、14h分别采集静脉血样约4mL，置于含肝素钠抗凝剂的真空采血管中。将全血样本置于-70℃冰箱中保存。
取全血样品100μL，加入40μL环孢素-d4工作溶液(内标，1.0μg/mL)混匀，加入异丙醇-乙腈(1∶2，v/v)溶液210μL，涡旋3min，4℃条件下以5000rpm转速离心15min，取上清液5μL注入HPLC-MS/MS系统进行分析。色谱条件为C18柱(2.0mm×50mm，5.0μm，Phenomenex)；流动相为10mmol/L醋酸铵溶液-乙腈＝7∶3(v/v)，等度洗脱，流速0.8mL/min，柱温40℃。质谱条件：ESI源，正离子多重反应监测模式，环孢素的检测离子对为1220.1→1203.2(m/z)，内标环孢素-d4的检测离子对为1225.1→1208.2(m/z)。
将获得的卡马西平片和环孢素软胶囊血药浓度数据采用WinNonlin 8.2传统药动学模式计算AIC值，以判断卡马西平和环孢素的房室模型；再以NCA模式分别计算两模型药物的Cmax、Tmax、AUC0-t、AUC0-∞和消除半衰期(t1/2)等药动学参数。
3.2.3 ka解析方法
3.2.3.1体内不依赖于房室模型的吸收动力学模型
分别将单室模型和双室模型设定参数值后所得的血药浓度数据、卡马西平片和环孢素软胶囊的临床药动学数据，输入至Python迭代法程序中(附录A)，ka取值范围为0.01～10.0h-1，k取值范围为0.01～5.0h-1，运行程序即可获得ka值。
3.2.3.2 Wagner-Nelson法
Wagner-Nelson法用于单室模型参数设定组及满足单室模型的临床药动学数据(卡马西平片)的ka值计算，以作为不依赖于房室模型的吸收动力学模型的对比研究。具体计算过程如下：

ln(1-Fabs)＝-kat+b   (公式10)
其中，Fabs为药物体内吸收分数；(XA)t和(XA)∞分别表示t时刻和t无穷大时进入体循环的药物量；Ct表示t时刻的血药浓度。所以，将Fabs与t进行线性回归，求得直线方程，其斜率即为ka值(公式10)。
3.2.3.3 Loo-Riegelman法
Loo-Riegelman法用于双室模型参数设定组及满足双室模型的临床药动学数据(环孢素软胶囊)的ka值计算，以作为不依赖于房室模型的吸收动力学模型的对比研究。具体计算过程如下：


ln(1-Fabs)＝-kat+b   (公式13)
其中，(Xp)t/Vc表示在t时刻进入外周室的药物量。Δc和Δt分别代表两个连续样本之间的血药浓度差值和时间间隔。所以，将Fabs与t进行线性回归，求得直线方程，其斜率即为ka值(公式13)。
3.2.3.4统计矩法
本研究将统计矩法应用于双室模型参数设定组的ka值计算，以作为不依赖于房室模型的吸收动力学模型和Loo-Riegelman法的对比研究。具体计算过程如下：

其中，Ci、Ci+1和Cn分别表示ti、ti+1和tn时间点的药物浓度；MAT为平均吸收时间；MRT为体内药物平均驻留时间；kT为终端消除速率常数；AUMC代表时间-血药浓度的乘积与时间的曲线下面积。
3.2.4 ka准确性的验证方法
对于单室模型和双室模型的参数设定组，ka的设定值即为真实值。采用不依赖于房室模型的吸收动力学模型、Wagner-Nelson法、Loo-Riegelman法或统计矩法估算ka值的准确性以RE表示，如公式4所示：
对于临床试验数据，由于无法获取客观准确的各模型药物ka值，本研究将采用反映药物吸收特征的药动学参数(Cmax、Tmax、Cmax/AUC0-t)与ka值进行Pearson相关性分析(SPSS 25.0，SPSS Inc.)，间接验证ka值的准确性。
3.2.5统计学方法
所有数据均表示为平均值±标准差。采用SPSS 25.0软件中的两独立样本t检验对数据进行统计学分析，p＜0.05表示两组具有显著性差异。
3.3结果与讨论
3.3.1房室模型设定参数验证不依赖于房室模型的吸收动力学模型及其解析方法的准确性
3.3.1.1单室模型参数变化
ka取值为0.15～5.00h-1(即吸收半衰期t1/2，abs为0.14～4.62h)，k取值为0.01～2.01h-1(即消除半衰期t1/2为0.34～69.30h)，满足了大多数单室模型药物的ka和k值范围。在不同的参数取值下，对比了不依赖于房室模型的吸收动力学模型的方法和Wagner-Nelson法计算ka值的准确性。
结果如图1所示，当k＝0.10h-1不变，ka设定值由0.15～5.00h-1变化时，药时曲线采用不依赖于房室模型的吸收动力学模型的方法计算所得ka值与ka真实值基本一致，Wagner-Nelson法所得ka值的准确性随着ka设定值的增大而稍有降低，但均在100±15％的准确度范围内；当ka＝3.00h-1不变，k设定值由0.01 ～2.01h-1变化时，药时曲线采用不依赖于房室模型的吸收动力学模型计算所得ka值与ka真实值基本一致，但Wagner-Nelson法估算ka值的准确性随着k设定值的增大而逐渐降低，当k取值为1.5h-1以上时，Wagner-Nelson法的准确度已低于85％。
结果表明，与Wagner-Nelson法相比，体内吸收动力学不依赖于房室模型的吸收动力学模型估算ka值的准确性更高，且准确性不受k和ka取值的影响。所以，不依赖于房室模型的吸收动力学模型估算ka的准确性与单室模型参数取值无关(即不依赖于房室模型参数)，仅与药时曲线的形状有关。客观而言，不依赖于房室模型的吸收动力学模型估算单室模型药物ka值的准确度理应较高，因为当单室模型的V、F、X0设置为固定值后，其血药浓度C与时间t关系式与吸收动力学不依赖于房室模型的吸收动力学模型的C-t关系式(公式1)基本一致。
3.3.1.2双室模型参数变化
从文献中获取了36个不同药物制剂的人体内血药浓度数据，采用WinNonlin软件计算各药物的AIC值。结果如表1所示，所有药物的AIC2值(双室模型)均小于AIC1值(单室模型)，表明36个药物的体内过程均符合双室模型。再通过WinNonlin软件初步估算了各药物的ka(0.210～1.826h-1)、k12(0.044～0.847h-1)、k21(0.010～0.451h-1)和k10(0.012～1.003h-1)范围。通过对比各参数间的关系发现，所有药物的k12和k10值之和均小于ka值(即ka＞k12+k10)，各药物的ka值均大于k12值，且k12值均高于k21值(即ka＞k12＞k21)。除少数药物(如阿昔洛韦、达卡他韦和左炔诺孕酮)外，k10值均显著高于k21(p＜0.05)，该结果为满足双室模型药物的ka、k10、k12和k21参数设定提供重要依据。
表1文献中获取36个药物制剂的人体内血药浓度数据及解析药动学参数
(***p＜0.001为ka对比k12、k21和k10；**p＜0.01为k12对比k21；*p＜0.05为k10对比k21)。


为了考察不依赖于房室模型的吸收动力学模型解析双室模型药物ka值的准确性与灵敏度，基于表1中36个不同药物的ka、k12、k21、k10值，按照降序排列，分别获取高、中、低三个水平的参数值，其中ka为1.098、0.603、0.375h-1；k12为0.525、0.211、0.133h-1；k21为0.176、0.067、0.025h-1；k10为0.571、0.271和0.100h-1。值得注意的是，通过获取文献中的药动学参数以降低设置组数和避免无效数据组，而非限定各参数的适用范围。根据ka、k12、k21、k10之间的关系(即ka＞k12+k10，且ka＞k12＞k21)，随机组合ka、k12、k21和k10的数值，共得到39组数据，并根据公式6～公式8计算每组的药时曲线(图2)。39组数据所对应的药时曲线均符合双室模型(AIC1＞AIC2，表2)。再分别通过不依赖于房室模型的吸收动力学模型的方法、Loo-Riegelman法和统计矩法估算各组的ka值。
结果如表2所示，采用不依赖于房室模型的吸收动力学模型估算的ka值与ka真实值相比，RE有正值和负值，所有RE值均在±16％以内，其中大部分RE值在±10％以内，表明不依赖于房室模型的吸收动力学模型解析双室模型药物ka值的准确度较高。采用Loo-Riegelman法估算的ka值与真实值相比，RE值变化幅度较大，且均为正值(即估算ka＞真实ka)，而采用统计矩法计算ka的大部分RE为负值(即估算ka＜真实ka)，准确性较差。由于统计矩法受到终端消除速率常数的影响，导致部分数据组的MAT为负值而无法获得ka值。
表2采用不同方法估算的双室模型参数设置组ka值的准确性



附注：aNA：MAT为负值，无法计算。
将表2中三种方法估算ka的RE绝对值进行汇总，采用不依赖于房室模型的吸收动力学模型计算ka的RE绝对值显著小于Loo-Riegelman法(p＜0.001)和统计矩法(p＜0.001)，但Loo-Riegelman法计算ka的准确性显著高于统计矩法(p＜0.01)，这与文献报道的结果一致。不依赖于房室模型的吸收动力学模型估算ka的RE中位数为-1.64％，与Loo-Riegelman法(21.5％)和统计矩法(-65.9％)相比，更接近于0。 另外，使用不依赖于房室模型的吸收动力学模型估算ka值不受k12、k21和k10等参数变化的影响，均保持较好的准确性，表明了不依赖于房室模型的吸收动力学模型不依赖于房室模型参数；由于Loo-Riegelman法为双室模型的经典方法，该方法计算过程中使用k12、k21和k10等参数，所以这些参数的取值变化对ka准确度较为敏感，但仍比统计矩法有更好的准确性；统计矩法为非房室模型方法，几乎不受k12、k21和k10变化的影响，但准确度较差。因此，以上结果证明了不依赖于房室模型的吸收动力学模型不依赖于体内双室模型参数(k12、k21、k10等)，而无需静脉血药浓度数据，该方法具有较高的准确性，满足不同类型药物的ka值解析。
3.3.2临床试验数据验证不依赖于房室模型的吸收动力学模型及其解析方法的准确性
3.3.2.1卡马西平片和环孢素软胶囊的ka值
卡马西平与环孢素均为窄治疗窗药物，所以临床BE试验为两序列、四周期自身交叉试验设计。在空腹和餐后状态下，卡马西平片和环孢素软胶囊口服给药后的人体内药时曲线见图3，药动学参数汇总于表3。
表3卡马西平片和环孢素软胶囊的人体内药动学参数

(平均值±标准差，卡马西平空腹和餐后n＝60，环孢素空腹n＝46，环孢素餐后n＝90，**p＜0.01为相同药物制剂的餐后与空腹药动学参数比较)。
卡马西平片空腹状态下参比制剂平均药时曲线的AIC1(-71.19)＜AIC2(34.78)，且餐后状态下参比制剂平均药时曲线的AIC1(-73.83)＜AIC2(35.10)，均符合单室模型。参比制剂与受试制剂在空腹和餐后状态下的药时曲线较为接近。空腹给药后，两制剂的Tmax约为3.0h，而餐后达峰时间延后至4.9h。与空腹相比，餐后两制剂的Cmax稍有降低，其余药动学参数(AUC0-t、AUC0-∞、t1/2)无明显变化，所以卡马西平片体内暴露量受食物影响较小。
环孢素软胶囊空腹状态下参比制剂平均药时曲线的AIC1(113.11)＞AIC2(51.13)，且餐后状态下参比制剂平均药时曲线的AIC1(89.99)＞AIC2(55.84)，均符合双室模型。参比制剂与受试制剂在空腹和餐后状态下的药时曲线较为接近。空腹给药后，两制剂的Tmax约为1.3h，而餐后达峰时间延后至2.5h。与空腹相比，餐后两制剂的Cmax降低了约2倍，AUC0-t、AUC0-∞降低了约1.4倍，t1/2无明显变化，表明食物对环孢素软胶囊的体内药动学特征具有显著影响。此外，由于环孢素软胶囊餐后的个体内变异较大，所以适当增加了受试者数量。
由于卡马西平片的药时曲线符合单室模型，分别采用不依赖于房室模型的吸收动力学模型的方法和Wagner-Nelson法估算卡马西平参比制剂与受试制剂在空腹和餐后状态下的ka值。结果显示(表4)，不依赖于房室模型的吸收动力学模型的方法计算的卡马西平两制剂的ka值在相同状态下几乎无差异，餐后状态下的ka均值略低于空腹状态，但无显著性差异。Wagner-Nelson法计算的卡马西平参比制剂和受试制剂及所有状态下的ka值均基本一致。
表4不依赖于房室模型的吸收动力学模型方法和Wagner-Nelson法计算卡马西平片的ka值

(平均值±标准差，n＝60)。
表5不依赖于房室模型的吸收动力学模型方法和Loo-Riegelman法计算环孢素软胶囊的ka值
(平均值±标准差，空腹n＝46，餐后n＝90，***p＜0.001为同一方法检测相同药物制剂的餐后与空腹ka值比较)。
由于环孢素软胶囊的药时曲线符合双室模型，分别采用不依赖于房室模型的吸收动力学模型的方法和Loo-Riegelman法计算环孢素软胶囊参比制剂与受试制剂在空腹和餐后状态下的ka值。其中，Loo-Riegelman法所需的k10、k12、k21参数值从文献中静脉血药浓度解析获取；此外，由于统计矩法的准确度较差，因此未被应用于环孢素软胶囊的ka解析。不同方法的ka解析结果如表5所示，不依赖于房室模型的吸收动力学模型的方法和Loo-Riegelman法估算的环孢素参比制剂与受试制剂的ka值在相同状态下无显著差异，由于受到食物影响，餐后状态下的两制剂ka值均显著低于空腹状态下的ka值(p＜0.001)。
3.3.2.2 ka值准确性的间接验证
已有报道显示，药物的体内吸收速率越快，导致血药浓度的Cmax越高，且Tmax缩短。Cmax和Cmax/AUC0-t值代表了药物的体内暴露量，与ka值的大小密切相关。由于无法获取卡马西平和环孢素的ka客观真实值，本研究通过将不同方法计算的ka值与Tmax、Cmax、Cmax/AUC0-t等反映药物吸收特征的药动学参数进行相关性分析，间接验证ka的准确性。对于卡马西平片，不依赖于房室模型的吸收动力学模型计算的ka值与Tmax呈负相关性(R＝-0.999，p＜0.01)，与Cmax呈正相关性(R＝0.983，p＜0.05)，与Cmax/AUC0-t具有潜在的正相关性(R＝0.949，p＝0.051)；使用Wagner-Nelson法估算的ka值与Cmax呈正相关性(R＝0.977，p＜0.05)，但与Tmax和Cmax/AUC0-t的相关性相对较差(p＞0.05)。结果表明，不依赖于房室模型的吸收动力学模型与Wagner-Nelson解析的卡马西平片ka值均具有良好的准确性，且不依赖于房室模型的吸收动力学模型的方法的准确性稍优于Wagner-Nelson法。
对于环孢素软胶囊，不依赖于房室模型的吸收动力学模型计算的ka值与Tmax呈负相关性(R＝-0.979，p＜0.01)，与Cmax和Cmax/AUC0-t均呈正相关性(R＞0.98，p＜0.05)；采用Loo-Riegelman法估算的ka值与Tmax、Cmax、Cmax/AUC0-t的相关系数R＞0.93，但ka值与三个药动学参数的p值略高于0.05，为潜在的相关性。结果表明，不依赖于房室模型的吸收动力学模型与Loo-Riegelman计算的环孢素软胶囊ka值均具有良好的准确性，且不依赖于房室模型的吸收动力学模型的方法的准确性优于Loo-Riegelman法。
3.3.2.3卡马西平片和环孢素软胶囊的Cmax预测
在不依赖于房室模型的吸收动力学模型(Python程序)的运行过程中，通过ka与k的迭代取值以拟合药时曲线，直至与实测药时曲线的残差绝对值之和最小。若拟合的药时曲线吸收相和Cmax与实测血药浓度数据越接近，代表ka估算的准确性越高。
结果显示，采用不依赖于房室模型的吸收动力学模型的方法所得的卡马西平片两制剂在不同状态下的药时曲线拟合度均较好，基本与实测药时曲线重合(图4A～图4D)，且Cmax平均预测误差(PE％)均在±5％以内(表6)，准确度高；环孢素软胶囊两制剂在不同状态下的药时曲线吸收相拟合度均较好(图4E～图4H)，由于环孢素为双室模型药物，处置相拟合相对较差，但不影响ka的计算。Cmax平均预测误差均在10％以内(表6)，准确度较好。因此，不依赖于房室模型的吸收动力学模型可以准确预测体内药物的吸收相和Cmax，进一步验证了ka解析的准确性。
表6不依赖于房室模型的吸收动力学模型的方法预测卡马西平片和环孢素软胶囊Cmax的误差

注：
4结论
本发明分别通过单、双房室模型参数设值和临床实测数据验证不依赖于房室模型的吸收动力学模型及其解析方法的准确性。房室模型参数验证结果显示，不依赖于房室模型的吸收动力学模型解析ka值的准确性略优于Wagner-Nelson，优于Loo-Riegelman和统计矩法，且计算过程及其准确性与各房室模型参数无关。临床结果显示，不依赖于房室模型的吸收动力学模型成功应用于卡马西平片和环孢素软胶囊两个模型药物的ka解析，且与反映药物体内吸收特性的药动学参数(Cmax、Tmax等)均具有良好的相关性，还可准确预测药物的吸收相与Cmax。因此，本申请提供了不依赖于房室模型的ka解析不依赖于房室模型的吸收动力学模型及方法，且准确度较高，适用范围更广。该模型为药物制剂的体内吸收动力学解析及其IVIVC的体内评价提供重要的工具。
关于体内吸收速率常数的解析方法(Python迭代法程序)。



Claims (10)

  1. 一种药物吸收速率常数的预测模型，其特征在于，所述模型包括：
    获取模块，用于获取药物制剂的实测血药浓度、时间采样点，进而绘制实测药时曲线，获得药物体内吸收特性的药动学参数；
    拟合模块，根据实测药时曲线、Tmax，输出不依赖于房室模型的吸收动力学模型，并获得估算ka值；
    真实模块，根据血药浓度特征选择房室模型预输入函数、实测血药浓度、时间采样点、固定参数，得到真实ka值；
    比对模块，对比估算ka值与真实ka值的差异，输出准确性高的估算ka值。
  2. 根据权利要求1所述预测模型，其特征在于，还包括修正模块，利用估算ka值预测药物制剂的吸收相与预测Cmax值，比对真实Cmax和预测Cmax，输出最佳估算ka值。
  3. 根据权利要求1所述预测模型，其特征在于，所述不依赖于房室模型的吸收动力学模型的建立过程如下：
    基于实测药时曲线，以Tmax对应的血药浓度点，将实测药时曲线分为血药浓度随时间的线性动力学上升和线性动力学下降两个过程，实测药时曲线的吸收相进行反卷积分获取药物体内吸收曲线，构建血药浓度与时间C-t的表达关系式，从而建立不依赖于房室模型的吸收动力学模型，其中，C-t表达关系式为：
    其中，ka为线性动力学上升速率常数；k为线性动力学下降速率常数；A为校正系数。
  4. 根据权利要求1所述预测模型，其特征在于，所述估算ka值解析过程如下：
    步骤1：基于药物制剂的实测血药浓度，血药浓度C经对数化后，取达峰后的连续若干数值，拟合线性回归，线性方程为LnC＝-kt+LnA(公式2)，其中截距为LnA，获得A值；
    步骤2：基于公式1和步骤1获得的A值，运用Python迭代法程序将k值与ka值按照设定的数值范围持续遍历带入公式1，并按照每0.01的数值进行迭代，以得到与实测药时曲线残差绝对值之和最小的预测药时曲线；
    步骤3：运用公式3计算步骤2获得的药时曲线中药物制剂的预估血药浓度与实测血药浓度的残差绝对值之和最小，进而输出最佳ka值；
    其中，Ci为实测的血药浓度，Ci’为估算的血药浓度。
  5. 根据权利要求1所述预测模型，其特征在于，所述真实ka值的获取过程如下：
    方式1：
    基于单室模型血管外给药的血药浓度计算公式5，
    将X0、F、V随机设置为固定值，分为两种情况：①保持k值不变，随机变化ka值；②保持ka值不变，随机变化k值；计算不同时间点的血药浓度，获得满足单室模型的多组特征药时曲线，获得真实ka值；
    方式2：
    基于双室模型血管外给药的血药浓度计算公式6～公式8，将X0、F、Vc随机设置为固定值；从现有技术中获取满足双室模型药物的血药浓度数据；使用WinNonlin软件初步计算这些药物的ka、k12、k21和k10值，并比较各参数间的关系；
    将所获得的ka、k12、k21和k10值按降序排序，每个参数的不同级别间随机组合，将这些参数带入公式6～公式8，进而获得不同时间点的血药浓度，绘制药时曲线；
    公式6中，α表示分布相混合一级速率常数，β表示消除相混合一级速率常数，分别由公式7、公式8计算；

    获得真实ka值。
  6. 根据权利要求5所述预测模型，其特征在于，方式1中，对单室模型来说，当k＝0.10h-1不变，ka设定值由0.15～5.00h-1变化时，药时曲线采用所述模型计算所得ka值与真实ka值基本一致；当ka＝3.00h-1不变，k设定值由0.01～2.01h-1变化时，药时曲线采用不依赖于房室模型的吸收动力学模型计算所得ka值与真实ka值基本一致；
    方式2中，对双室模型来说，基于满足双室模型药物的ka、k12、k21、k10值，随机组合ka、k12、k21 和k10的数值，并根据公式6～公式8计算每组的药时曲线；采用不依赖于房室模型的吸收动力学模型的估算ka值与真实ka值相比，所有RE值均在±16％以内。
  7. 根据权利要求5所述预测模型，其特征在于，所述满足双室模型药物为醋酸阿比特龙片、阿昔洛韦混悬剂、阿奇霉素片、苯那普利胶囊、安非他酮片、坎地沙坦酯片、卡托普利片、塞来昔布胶囊、环丙沙星片、氯吡格雷片、达卡他韦片、多潘立酮片、屈他维林片、格列本脲片、氢氯噻嗪片、伊拉地平胶囊、伊曲康唑片、拉西地平片、盐酸乐卡地平片、左炔诺孕酮片、氯雷他定片、二甲双胍片、吗替麦考酚片、萘普生片、奥美沙坦酯片、磷酸奥司他韦胶囊、喹那普利片、瑞格列奈片、利匹韦林片、瑞舒伐他汀片、西洛多辛胶囊、辛伐他汀片、替米沙坦片、富马酸替诺福韦酯片、特比萘芬片、替格瑞洛片。
  8. 一种药物吸收速率常数准确性的验证方法，其特征在于，运用权利要求1-7任一项所述模型基于现有房室模型参数设定和临床试验数据验证计算的吸收速率常数ka的准确性。
  9. 一种计算机设备，其特征在于，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机中权利要求1-7任一项所述模型。
  10. 一种计算机可读存储介质，其特征在于，所述可读存储介质存储有计算机程序，所述计算机程序被处理器执行实现权利要求1-7任一项所述模型的执行。
PCT/CN2023/098532 2022-06-14 2023-06-06 药物吸收速率常数的预测模型、设备和存储介质 WO2023241402A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210668645.7A CN114822876B (zh) 2022-06-14 2022-06-14 药物吸收速率常数的预测模型、设备和存储介质
CN202210668645.7 2022-06-14

Publications (1)

Publication Number Publication Date
WO2023241402A1 true WO2023241402A1 (zh) 2023-12-21

Family

ID=82521691

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/098532 WO2023241402A1 (zh) 2022-06-14 2023-06-06 药物吸收速率常数的预测模型、设备和存储介质

Country Status (2)

Country Link
CN (1) CN114822876B (zh)
WO (1) WO2023241402A1 (zh)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114822876B (zh) * 2022-06-14 2023-07-21 湖南慧泽生物医药科技有限公司 药物吸收速率常数的预测模型、设备和存储介质
CN115270524B (zh) * 2022-09-27 2023-01-17 天津天堰科技股份有限公司 一种模拟bis的方法
CN115598262B (zh) * 2022-11-24 2023-03-03 则正(济南)生物科技有限公司 一种双氯芬酸钠缓释片体内外相关性评价的方法
CN115565621B (zh) * 2022-11-29 2023-03-31 则正(济南)生物科技有限公司 茶碱缓释片体内外相关性模型、构建方法及应用

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060120962A1 (en) * 2004-10-12 2006-06-08 Rabinowitz Joshua D Cardiac safe, rapid medication delivery
CN105044318A (zh) * 2015-09-22 2015-11-11 河南科技大学 一种家兔组织中兽药残留浓度的预测方法
CN109285588A (zh) * 2018-09-05 2019-01-29 东南大学 大鼠壬基酚的生理毒代动力学模型构建方法
CN110911016A (zh) * 2019-12-12 2020-03-24 南京航空航天大学 基于非房室模型的血浆及组织的药代动力学参数计算方法
CN113035369A (zh) * 2021-03-10 2021-06-25 浙江大学 一种肾移植抗感染药物剂量预测模型的构建方法
CN114822876A (zh) * 2022-06-14 2022-07-29 湖南慧泽生物医药科技有限公司 药物吸收速率常数的预测模型、设备和存储介质

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1811975A2 (en) * 2004-10-19 2007-08-01 The State of Oregon Acting by and Enteric coated compositions that release active ingredient(s) in gastric fluid and intestinal fluid
CN103077296B (zh) * 2011-10-25 2015-11-18 复旦大学附属华山医院 基于流速调节的静脉注射药动学模型的体外模拟方法
CN103018414A (zh) * 2012-10-25 2013-04-03 中南大学 一种药物制剂活性成分个性化体外溶出检测方法和溶出装置
CN106126963B (zh) * 2016-08-18 2018-10-19 南京诺尔曼生物技术有限公司 一种模拟药-时曲线的方法
CN111983209A (zh) * 2019-05-22 2020-11-24 天士力医药集团股份有限公司 一种评价复方丹参滴丸多组分在大鼠体内暴露量的采样点优化方法
TW202221720A (zh) * 2020-08-03 2022-06-01 大陸商四川海思科製藥有限公司 確定異丙酚及其衍生物群體藥代動力學模型的方法及系統
CN113077846A (zh) * 2021-04-01 2021-07-06 深圳市药品检验研究院（深圳市医疗器械检测中心） 一种氯雷他定pbpk模型的建立方法及其应用
CN113990519A (zh) * 2021-10-19 2022-01-28 药融云数字科技(成都)有限公司 一种pbpk-ro模型预测塞尔帕替尼联合用药后体内pk/pd变化的方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060120962A1 (en) * 2004-10-12 2006-06-08 Rabinowitz Joshua D Cardiac safe, rapid medication delivery
CN105044318A (zh) * 2015-09-22 2015-11-11 河南科技大学 一种家兔组织中兽药残留浓度的预测方法
CN109285588A (zh) * 2018-09-05 2019-01-29 东南大学 大鼠壬基酚的生理毒代动力学模型构建方法
CN110911016A (zh) * 2019-12-12 2020-03-24 南京航空航天大学 基于非房室模型的血浆及组织的药代动力学参数计算方法
CN113035369A (zh) * 2021-03-10 2021-06-25 浙江大学 一种肾移植抗感染药物剂量预测模型的构建方法
CN114822876A (zh) * 2022-06-14 2022-07-29 湖南慧泽生物医药科技有限公司 药物吸收速率常数的预测模型、设备和存储介质

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LING QIU, HUANG XIONGBO; LIU WUPING: "Development and application of pharmacokinetic system of optimal compartmental model", CHINESE JOURNAL OF CLINICAL RATIONAL DRUG USE, vol. 14, no. 10A, 10 October 2021 (2021-10-10), pages 1 - 4, 7, XP093118253 *
ZHENG LI, XIAOLI YU: "Exploring frequency differences of physiological processes to enhance dynamic FDG-PET without blood function", 2003 IEEE NUCLEAR SCIENCE SYMPOSIUM. CONFERENCE RECORD (IEEE CAT. NO.03CH37515), IEEE, vol. 4, 1 January 2003 (2003-01-01), pages 2844 - 2848, XP093118248, ISSN: 1082-3654, ISBN: 978-0-7803-8257-2, DOI: 10.1109/NSSMIC.2003.1352477 *

Also Published As

Publication number Publication date
CN114822876B (zh) 2023-07-21
CN114822876A (zh) 2022-07-29

Similar Documents

Publication Publication Date Title
WO2023241402A1 (zh) 药物吸收速率常数的预测模型、设备和存储介质
Salman et al. Pharmacokinetic comparison of two piperaquine-containing artemisinin combination therapies in Papua New Guinean children with uncomplicated malaria
De Kock et al. Pharmacokinetics of sulfadoxine and pyrimethamine for intermittent preventive treatment of malaria during pregnancy and after delivery
Hobl et al. A randomized, double-blind, parallel, single-site pilot trial to compare two different starting doses of methotrexate in methotrexate-naïve adult patients with rheumatoid arthritis
Kamel et al. Determination of febuxostat in human plasma by high performance liquid chromatography (HPLC) with fluorescence-detection
CN104483407B (zh) 血样中微量伊马替尼测定方法及在零期临床试验中的应用
CN102253129A (zh) 一种同时测定多种抗hiv药物血浆浓度方法
Mehta et al. Pharmacokinetics of fluticasone propionate and salmeterol delivered as a combination dry powder via a capsule-based inhaler and a multi-dose inhaler
Tao et al. Bioequivalence and food effect assessment of 2 rivaroxaban formulations in healthy Chinese volunteers: An open, randomized, single‐dose, and 4‐period crossover study
CN115240875B (zh) 一种药物吸收速率常数的直接估测方法
Tian et al. Effects of plasma albumin on the pharmacokinetics of esomeprazole in ICU patients
Caballero et al. Eosinophil count in induced sputum could be more sensitive than in peripheral blood to phenotype patients with severe eosinophilic asthma
Zhang et al. Pharmacokinetics, pharmacodynamics, tolerability and safety of single doses of bivalirudin in healthy Chinese subjects
Liang et al. Population pharmacokinetic analysis and dosing optimization of polymyxin B in critically ill patients
Pawinski et al. A limited sampling strategy for estimating mycophenolic acid area under the curve in adult heart transplant patients treated with concomitant cyclosporine
Kim et al. Pharmacokinetics of lacosamide in healthy Korean male volunteers
WO2014059197A1 (en) Methods and compositions for diagnosing and treating hyposmia
Wang et al. Investigation of a potential drug-drug interaction between salbutamol and ambroxol and bioequivalence of a new fixed-dose combination containing these two drugs in healthy Chinese subjects
Noh et al. Pharmacokinetic comparison of 2 formulations of anastrozole (1 mg) in healthy Korean male volunteers: a randomized, single-dose, 2-period, 2-sequence, crossover study
Główka et al. Bioavailability of propylthiouracil from two formulation tablets
CN112730682B (zh) 一种抗病毒药物临床研究血浆样本中奥司他韦及代谢物奥司他韦酸浓度的生物分析方法
Mendoza-Tamayo et al. Bioavailability of two tablet formulations of a single dose of moxifloxacin 400 mg: An open-label, randomized, two-period crossover comparison in healthy Mexican adult volunteers
Chandra et al. LC-MS/MS assay for the determination of apixaban in human plasma: its application to a pharmacokinetic study
Aljuffali et al. Pharmacokinetic assessment of ketanserin in the horse
Chowdhury et al. Pharmacokinetic parameters of ciprofloxacin in Bangladeshi volunteers: A preliminary evaluation

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23822971

Country of ref document: EP

Kind code of ref document: A1