Blood coagulation factor drug delivery model for perioperative period A of hemophilia and application of blood coagulation factor drug delivery model

Abstract

The invention relates to a hemophilia A perioperative period blood coagulation factor drug delivery model and application thereof, and relates to the technical field of biological information. The hemophilia A perioperative period blood coagulation factor administration model is constructed by adopting a nonlinear mixed effect modeling method based on a two-chamber model structure, and covariants of the hemophilia A perioperative period blood coagulation factor administration model comprise blood type, body weight, age, perioperative period anti-fibrinolytic drug use condition and perioperative period NSAID analgesic drug use condition NSAID. The model can be used for predicting an FVIII concentration-time curve of a clinical individual patient, and individual fine optimization of the administration dosage and the administration time is realized.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics, and in particular to a perioperative coagulation factor administration model for hemophilia A and its application. Background Technology

[0002] Currently, perioperative coagulation factor replacement therapy for hemophilia A mainly employs two dosing regimens: one is to calculate the dosage based on the patient's weight using an empirical formula (weight-based method), and the other is to obtain individual parameters such as half-life and clearance rate through preoperative pharmacokinetic studies to develop an individualized dosing regimen (preoperative PK method). The traditional weight-based method is widely used in clinical practice, is simple to operate, and relies on guideline-recommended standard doses. However, it ignores the differences in important pharmacokinetic parameters such as clearance rate, volume of distribution, and half-life of coagulation factors among patients, which can easily lead to insufficient or excessive dosage, increasing the risk of perioperative bleeding or treatment costs.

[0003] Preoperative PK method involves multiple blood samplings after intravenous infusion of the test dose to model individual parameters, and uses population pharmacokinetic model and Bayesian estimation method to fit individual parameters and formulate personalized alternatives accordingly, which improves the accuracy and economy of factor use to a certain extent. However, the preoperative PK method still has the following defects: (1) The test design is complex and time-consuming, requiring multiple intravenous blood samplings, which is a heavy burden on patients; (2) The obtained PK parameters are based on the preoperative resting state and are difficult to reflect the actual changes in coagulation factor metabolism under intraoperative and postoperative stress; (3) The implementation threshold is high, relying on specialized software and professional teams, which limits its promotion in ordinary medical institutions; (4) Some patients experience significant changes in their condition during the operation, resulting in distortion of the preoperative PK parameters and a large deviation between the predicted dose and the actual needs.

[0004] The fundamental reason for these limitations is that existing pharmacokinetic models fail to establish specific pharmacokinetic models for perioperative physiological states, resulting in large fluctuations in intraoperative coagulation factor levels and a lack of dynamic adjustment capabilities. Therefore, there is an urgent clinical need for an individualized dosing model that can reflect the actual perioperative physiological state and has higher predictive accuracy to optimize coagulation factor management in patients undergoing hemophilia A surgery. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a perioperative coagulation factor administration model for hemophilia A. This model can be used to predict the FVIII concentration-time curve for individual clinical patients, enabling individualized and precise optimization of dosage and administration time.

[0006] To achieve the above objectives, the present invention provides a perioperative coagulation factor administration model for hemophilia A. This perioperative coagulation factor administration model for hemophilia A is constructed based on a two-compartment model structure using a nonlinear mixed-effects modeling method. The covariates of the perioperative coagulation factor administration model for hemophilia A include blood type, weight, age, perioperative use of antifibrinolytic drugs, and perioperative use of NSAID analgesics.

[0007] The aforementioned model employs a nonlinear mixed-effects modeling (NLME) approach, establishing pharmacokinetic equations based on a two-compartment model structure. Internal model validation is performed using bootstrap and visual prediction test (VPC), and external validation is conducted on multi-center real-world data. The final model can be used for predicting FVIII concentration-time curves in individual clinical patients, enabling personalized and precise optimization of dosage and administration time.

[0008] In one embodiment, the use of perioperative antifibrinolytic drugs includes: use of antifibrinolytic drugs during the perioperative period or no use of antifibrinolytic drugs during the perioperative period; the use of perioperative NSAID analgesics includes: use of NSAID analgesics during the perioperative period or no use of NSAID analgesics during the perioperative period.

[0009] In one embodiment, the perioperative coagulation factor administration model for hemophilia A is shown below: V 2873.47 (0.28 )

[0010] 4093.37

[0011] 192.21

[0012]

[0013] Baseline value 1.09

[0014] Where V: central compartment distribution volume, unit: mL, representing the initial distribution volume of the drug; I Antifibrinolytic drug Perioperative antifibrinolytic drug use: A value of 1 indicates that antifibrinolytic drugs were used during the perioperative period, and a value of 0 indicates that antifibrinolytic drugs were not used during the perioperative period. V2: Peripheral compartment distribution volume, unit: mL, representing the volume of drug distributed in peripheral tissues; CL: Central compartment clearance rate, unit: mL / h, represents the volume of drug cleared from the central compartment per unit time, which is affected by individual characteristics and covariates; I Blood Type The patient's blood type: a value of 1 indicates that the patient's blood type is O, and a value of 0 indicates that the patient's blood type is not O. CL2: Distribution clearance rate of peripheral compartments, unit: mL / h, representing the exchange rate between the central compartment and the peripheral compartment; I Antiinflammatory drug The value of 1 indicates that NSAID analgesics were used during the perioperative period, and the value of 0 indicates that NSAID analgesics were not used during the perioperative period. Baseline value: Peak activity of coagulation factor VIII after preoperative loading dose, representing the initial concentration before surgery; η represents the inter-individual variation corresponding to each parameter.

[0015] When antifibrinolytic drugs are used during the perioperative period =1, exp(0.28) =1.323; when no antifibrinolytic drugs were used during the perioperative period, =0, exp(0.28) ) = 1; When the patient has type O blood = 0, exp( =1; when the patient's blood type is not O, = 1, exp( =exp() =0.787; When NSAID analgesics are used during the perioperative period =1,exp() =exp() =0.878; When no NSAID analgesics were used during the perioperative period, =0, exp( =1.

[0016] The research team of the inventors of this invention collected perioperative clinical data from patients with hemophilia A undergoing orthopedic surgery. Combined with multi-timepoint activity monitoring results of coagulation factor VIII (FVIII), they constructed a population pharmacokinetics (PPK) model reflecting the metabolic patterns of FVIII under perioperative physiological stress. Multiple individualized covariates were introduced to correct for parameter variability, resulting in the aforementioned perioperative coagulation factor administration model for hemophilia A. This perioperative coagulation factor administration model for hemophilia A addresses the problems of existing weight-based dosing methods ignoring individual differences and preoperative PK parameter guidance schemes failing to adapt to dynamic changes during the perioperative period.

[0017] This invention also provides the application of the perioperative coagulation factor administration model for hemophilia A in drug administration guidance, which includes the following steps: (1) Obtain the patient’s perioperative information, which includes coagulation factor administration scenario information and covariates of the hemophilia A perioperative coagulation factor administration model. The initial coagulation factor VIII administration regimen under the administration scenario is obtained by Monte Carlo random simulation. (2) After administration of the initial coagulation factor VIII according to the prescribed administration regimen, the peak activity of coagulation factor VIII after the preoperative loading dose is measured to assess whether the therapeutic window range has been reached. (3) Input the peak activity of coagulation factor VIII after the preoperative loading dose, the dosing information in the initial coagulation factor VIII dosing regimen, and the covariates of the hemophilia A perioperative coagulation factor dosing model into the hemophilia A perioperative coagulation factor dosing model, and estimate the individualized pharmacokinetic parameters of the patient by combining the maximum a posteriori Bayes method with prior information and observation data to obtain the individualized dosing regimen. (4) After administration of the individualized dosing regimen, perioperative coagulation factor VIII administration monitoring is performed to obtain the peak activity of coagulation factor VIII after the patient’s preoperative loading dose. Combined with the perioperative coagulation factor VIII activity treatment window, the individualized dosing regimen is given according to step (3). (5) Repeat steps (1)-(4) to ensure that the peak activity of coagulation factor VIII after the patient’s preoperative loading dose is within the therapeutic window at different stages of the perioperative period, thereby reducing the risk of perioperative bleeding.

[0018] In one embodiment, step (1) of the application, the initial coagulation factor VIII dosing regimen calculated by Monte Carlo random simulation, includes: importing the patient's covariate information and the data files of different dosing regimen scenarios to be simulated into Phoenix NMLE software, performing Monte Carlo random simulation, calculating the probability of achieving the perioperative coagulation factor concentration treatment target (e.g., the probability of activity above 80%) under different dosing regimens, trying multiple times, and selecting the dosing regimen with the highest probability of achieving the target as the final recommended dosing regimen.

[0019] Understandably, the data files for the different dosing methods mentioned above include the loading dose and the dosing regimen after the loading dose; for example, 1000 IU q 12h 8 hours after the loading dose, or 1000 IU q 8h 12 hours after the loading dose.

[0020] In one embodiment, step (2) of the application further includes: closely monitoring the patient's postoperative bleeding risk and wound healing.

[0021] In one embodiment, in step (3) of the application, the dosing information in the initial coagulation factor VIII dosing regimen is the drug usage and drug dosage in the initial coagulation factor VIII dosing regimen.

[0022] In one embodiment, in step (3) of the application, the prior information is the pharmacokinetic characteristic parameters of the target population, and the pharmacokinetic characteristic parameters include: the mean of the PPK parameter, the median of the PPK parameter, the inter-individual variation of the parameter, the intra-individual variation of the parameter and / or the factors that cause the variation of the parameter; the observation data includes the activity monitored after administration of coagulation factor VIII in the perioperative period.

[0023] Understandably, in step (5) of the application, without obtaining postoperative coagulation factor VIII concentration data, Bayesian feedback cannot be used to predict the recommended dosage regimen for subsequent concentrations. Therefore, Monte Carlo random simulation is used for estimation. Thus, when the patient is admitted to the hospital but has not yet undergone surgery, a postoperative first-day medication regimen is suggested through Monte Carlo simulation. Once postoperative coagulation factor VIII concentration data is obtained, Bayesian feedback is used to predict and obtain a new medication adjustment regimen.

[0024] In one embodiment, the treatment window range is the treatment window range recommended by the Chinese guidelines for the treatment of hemophilia.

[0025] The present invention also provides a system for providing a coagulation factor administration regimen, comprising: The data storage module is used to store patient information and the perioperative coagulation factor administration model for hemophilia A. A data analysis module is used to obtain individualized dosing regimens in steps (1)-(4) of the application; and The data display module is used to show patients' individualized medication regimens.

[0026] In one embodiment, the information includes: blood type, weight, age, use of antifibrinolytic drugs, and use of NSAID analgesics.

[0027] In one embodiment, the patient is a hemophilia A patient.

[0028] In one embodiment, the patient is in the perioperative period.

[0029] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses a perioperative coagulation factor administration model for hemophilia A and its application. This model addresses the problems of existing weight-based administration methods ignoring individual differences and preoperative PK parameter guidance schemes failing to adapt to dynamic perioperative changes. Based on real clinical data, this model significantly improves the accuracy and safety of perioperative FVIII replacement therapy. Compared to traditional weight-based or preoperative PK guidance methods, this invention's model has the following significant technical and clinical effects: 1. Improve the rate of coagulation factor activity reaching the target and reduce the risk of activity fluctuations: This model is based on bioactivity data from previous surgeries. Internal validation showed that 90% of the prediction interval covered most of the observations, and external validation showed that the F20 / F30 indices reached 35.0% and 53.3%, respectively, indicating good predictive accuracy. Exploratory trials showed that, under the guidance of the model, the postoperative FVIII bioactivity failure rate was controlled at 20.0%–33.3%, significantly better than the weight-based method (which had a failure rate of over 70%).

[0030] 2. Reduces the dosage of clotting factors and lowers the financial burden: The average factor usage in the three patients actually guided by the model was significantly lower than that in the preoperative PK group and the total body weight group. Factor costs were reduced by an average of more than 35%, which helps alleviate the economic burden caused by the high cost of factor use.

[0031] 3. Enhance intraoperative stress adaptability and achieve individualized dynamic management: The model considers multiple intraoperative covariates (such as age, blood type, perioperative use of antifibrinolytic drugs and perioperative use of NSAID analgesics), overcoming the limitations of the "static state" of the preoperative PK model. It can better reflect the actual needs during surgery and improve the stability of intraoperative factor level control, making it particularly suitable for cases with high trauma or large intraoperative fluctuations.

[0032] 4. Promote the integrated application of models in clinical information systems: This model can be embedded into hospital information systems or decision support platforms to automatically calculate and recommend coagulation factor dosing regimens, improve doctors' decision-making efficiency, and promote the implementation of personalized medicine.

[0033] 5. Significant social and public health benefits: Hemophilia patients face a high risk of perioperative bleeding and high treatment costs. This invention is expected to be promoted nationwide to improve surgical safety, reduce rebleeding and reoperation events, save medical insurance resources, and has broad social significance. Attached Figure Description

[0034] Figure 1 A and B are scatter plots of population predicted values ​​(PRED) and observed values ​​(DV) of the basic model and the final model, respectively, where A is the basic model and B is the final model. Figure 1 C and D are scatter plots of individual predicted values ​​(IPRED) and observed values ​​(DV) for the basic model and the final model, respectively, where C is the basic model and D is the final model; Figure 1 E and F are scatter plots of inter-individual variation predictions (IVAR) and conditionally weighted residuals (CWRES) for the basic model and the final model, respectively, where E is the basic model and F is the final model. Figure 2 G and H are scatter plots of population predictions and conditionally weighted residuals (CWRES) for the basic model and the final model, respectively, where G is the basic model and H is the final model. Figure 3 This is the final model VPC diagnostic diagram. Detailed Implementation

[0035] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0037] source: Unless otherwise specified, all reagents, materials, and equipment used in this embodiment are commercially available; unless otherwise specified, all test methods are conventional test methods in this field.

[0038] Example I. A perioperative coagulation factor administration model for hemophilia A.

[0039] This study successfully constructed a population pharmacokinetic (PPK) model for patients with hemophilia A undergoing orthopedic surgery during the perioperative period, as shown below: V 2873.47 (0.28 )

[0040] 4093.37

[0041] 192.21

[0042]

[0043] Baseline value 1.09

[0044] Where V: central compartment distribution volume, unit: mL, representing the initial distribution volume of the drug; I Antifibrinolytic drug Perioperative antifibrinolytic drug use: A value of 1 indicates that antifibrinolytic drugs were used during the perioperative period, and a value of 0 indicates that antifibrinolytic drugs were not used during the perioperative period. V2: Peripheral compartment distribution volume, unit: mL, representing the volume of drug distributed in peripheral tissues; CL: Central compartment clearance rate, unit: mL / h, represents the volume of drug cleared from the central compartment per unit time, which is affected by individual characteristics and covariates; I Blood Type The patient's blood type: a value of 1 indicates that the patient's blood type is O, and a value of 0 indicates that the patient's blood type is not O. CL2: Distribution clearance rate of peripheral compartments, unit: mL / h, representing the exchange rate between the central compartment and the peripheral compartment; I Antiinflammatory drug The value of 1 indicates that NSAID analgesics were used during the perioperative period, and the value of 0 indicates that NSAID analgesics were not used during the perioperative period. Baseline value: Peak activity of coagulation factor VIII after preoperative loading dose, representing the initial concentration before surgery; η represents the inter-individual variation corresponding to each parameter.

[0045] When antifibrinolytic drugs are used during the perioperative period =1, exp(0.28) =1.323; when no antifibrinolytic drugs were used during the perioperative period, =0, exp(0.28) ) = 1; When the patient has type O blood = 0, exp( =1; when the patient's blood type is not O, = 1, exp( =exp() =0.787; When NSAID analgesics are used during the perioperative period =1,exp() =exp() =0.878; When no NSAID analgesics were used during the perioperative period, =0, exp( =1.

[0046] This model employs a two-compartment structure and incorporates baseline FVIII level parameters, effectively describing the perioperative FVIII concentration-time dynamics. Compared to a one-compartment model, the two-compartment model significantly improves the fit, suggesting that FVIII is distributed in vivo through two compartments: a central compartment and a peripheral compartment. This is consistent with the results reported by Hazendonk et al. in the perioperative population. Typical model parameters are reasonably estimated, such as a clearance rate (Cl) of approximately 0.19 L / h (based on a 70 kg patient), a central volume of distribution of approximately 2.9 L, and an estimated elimination half-life of approximately 10–12 hours, consistent with the range of adult FVIII half-lives reported in previous literature. The accuracy of each fixed-effects parameter is high, with most coefficients of variation below 30%, and none crossing 0 in the 95% confidence intervals, indicating statistical significance and reliability of the model parameters. The model residual variance is also small (σ≈0.19), indicating limited deviation between observed and predicted values, and the model structure can adequately explain the pharmacokinetic characteristics of FVIII in this study population.

[0047] This study used a stepwise regression approach with a "broad entry, strict exit" method to screen covariates, ultimately including five covariates: blood type, weight, age, perioperative use of antifibrinolytic drugs, and perioperative use of NSAID analgesics. These covariates significantly affected the model parameters. First, blood type (O vs. non-O) was shown to significantly influence FVIII clearance. The model results showed that the clearance rate in non-O blood patients was approximately 24% lower than in O blood patients (θ_BloodType-Cl = –0.237), implying that FVIII clearance was faster in O blood patients. This trend is consistent with the study by Hazendonk et al., whose population model showed that O blood increased FVIII clearance by 26%. This phenomenon can be explained by the influence of blood type on vWF and FVIII metabolism: O blood patients have lower vWF levels, thus resulting in relatively faster FVIII clearance. Furthermore, the negative effect of age on FVIII clearance was also reflected in this model (θ_Age-Cl = –0.293), meaning that clearance rates were lower in older patients. Hazendonk et al. also observed a decrease in clearance rate with each increase in age, and a prolonged FVIII half-life in older adult patients. This suggests that changes in metabolic rate and vascular endothelial function with age may lead to a longer retention of FVIII in the body. Body weight, as an indicator of individual body size, has a positive impact on clearance rate (θ_Weight-Cl=0.437), indicating that greater body weight results in faster FVIII clearance. The influence of body weight on FVIII pharmacokinetics is also widely recognized: Hazendonk et al. used an allometric growth scale (body weight to the power of 0.75) to adjust for clearance and volume of distribution in their model; studies by Henrard and Hermans et al. further indicated that obese patients receive a higher increase in FVIII activity per kilogram of body weight (IVR value), while underweight patients have a lower IVR, suggesting that dosage calculations should refer to ideal body weight rather than actual body weight. Our model uses body weight as a covariate for clearance, which essentially takes into account the impact of individual body size differences on FVIII metabolism. This is consistent with the view that standard dosing based on actual body weight may produce biases in obese or emaciated patients, and body size correction is needed to improve the accuracy of individualized dosing.

[0048] Notably, this model incorporates perioperative adjuvant medication factors, which have not been reported in previous literature. The use of antifibrinolytic drugs (such as tranexamic acid) significantly affected the central distribution volume of FVIII (θ_Antifibrinolytic-V=0.276). This result may reflect the effect of antifibrinolytic therapy on perioperative coagulation dynamics: the use of antifibrinolytic drugs can reduce fibrin degradation and secondary fibrinolysis, and the hypercoagulable state may lead to a more widespread distribution of FVIII or alter its intravascular residence time, resulting in an increased apparent distribution volume. The use of NSAIDs was associated with a decrease in FVIII clearance (θ_Anti-inflammatory-Cl=–0.130), suggesting that perioperative use of certain NSAIDs may slow FVIII elimination. One possible explanation is that NSAID treatment reduces surgical stress and inflammatory response, thereby reducing the body's consumption or clearance of FVIII. However, due to the diversity of NSAIDs (such as selective COX-2 inhibitors used for analgesia), their specific mechanisms of action on FVIII pharmacokinetics remain unclear and require further investigation. This study incorporated the aforementioned medication factors into the model, improving its adaptability to different perioperative management protocols. In the model by Hazendonk et al., the impact of surgical risk level on clearance was considered (clearance rate decreased slightly by 7% in major surgeries, P<0.01). In our study, the surgical scale factor did not show significant results after incorporating the drug covariate. This may be because the use of antifibrinolytic drugs is often associated with major surgeries and high bleeding risks; therefore, the drug covariate in the model, to some extent, substitutes for the impact of surgical complexity on FVIII metabolism. This demonstrates the comprehensiveness of the covariates included in this model and also suggests that the potential impact of perioperative drug management on coagulation factor pharmacokinetics should be considered in clinical application.

[0049] II. Validation of the perioperative PK model.

[0050] 1. Goodness-of-fit plot.

[0051] To evaluate the model's fit and predictive ability, goodness-of-fit plots were created between the baseline model (which does not include covariates and only considers the relationship between coagulation factor usage and activity) and the final model (which includes covariates and baseline values). These plots included scatter plots of predicted and observed values, residual plots, and residual normality test plots. In the final model, the scatter points are more evenly and densely distributed near the diagonal, indicating improved predictive ability at both the population and individual levels. In the baseline model, CWRES showed a certain trend with IVAR, while in the final model, CWRES were approximately randomly distributed around the zero line within the IVAR range, suggesting no systematic bias in the error and a more reasonable model structure. The final model exhibited smaller residual fluctuations and lower dispersion, further validating its better stability and explanatory power at the population level.

[0052] Scatter plots of population predicted values ​​(PRED) and observed values ​​(DV) for the base model and final model are shown below. Figure 1 As shown in Figures A and B, the scatter plots of individual predicted values ​​(IPRED) and observed values ​​(DV) for the base model and the final model are as follows: Figure 1 As shown in C and D, the scatter plots of inter-individual variance predicted values ​​(IVAR) and conditionally weighted residuals (CWRES) for the basic model and the final model are as follows. Figure 1 As shown in E and F, the scatter plots of the population predictions and conditionally weighted residuals (CWRES) of the base model and the final model are as follows: Figure 2 G, H.

[0053] 2. Model parameter stability assessment (Bootstrap validation).

[0054] To verify the stability and reliability of the final pharmacokinetic model parameter estimation results, the Bootstrap method was used for internal validation. The sampling was repeated 1000 times and the mean, standard deviation (SD), coefficient of variation (CV%), median and 95% confidence interval (CI) of each parameter were calculated.

[0055] The final model estimates were compared with the Bootstrap analysis results. The Bootstrap mean values ​​of each structural parameter (such as tvV, tvCL, tvCL2, etc.) were highly close to the final model estimates, with deviations all within ±2.5%, and the relative deviations of most parameters were less than ±1%. For example, the estimated value of tvV was 2873.47, the Bootstrap mean was 2941.93, and the deviation was -1.209%; the estimated value of tvCL was almost identical to the Bootstrap mean (192.214 vs 192.338), with a relative deviation of 0.000%. These results show that the structural parameter estimates of the final model have good stability. None of the 95% confidence intervals for all parameters contained 0, indicating that these parameters are statistically significant. Furthermore, the CV% of the parameters were generally within a reasonable range, with the CV% of key parameters such as tvCL and tvV2 being below 30%, suggesting high estimation accuracy. The Bootstrap mean of the covariate parameter θ also showed good agreement with the final model estimates. Taking θ_BloodType-C1 as an example, its estimated value is -0.237, the Bootstrap mean is -0.238, and the 95% CI is (-0.335, -0.150), with a relative bias of only 0.152%. The effects of other covariates such as antifibrinolytic drugs, weight, age, and anti-inflammatory drugs on CL or V are all statistically significant (CIs do not cross zero), indicating that the inclusion of covariates in the model is reasonable and stable. Although the CV% of some covariates is relatively high (e.g., the CV% of weight's effect on CL is 29.452%), their 95% CIs are still relatively concentrated, supporting the reasonableness of their inclusion in the model. The inter-individual variation parameter (ω²) was supported by the Bootstrap. All parameters showed no anomalies in the Bootstrap results, and the means were close to the final estimates. For example, the final estimate of ω²_V2 is 1.230, and the Bootstrap mean is 1.394, indicating that the model's estimation of inter-individual variation is reproducible. Although the CV% of some IIV parameters is relatively high (e.g., CV% of ω²_V2 = 13.315%), it is still within an acceptable range, indicating that the estimation uncertainty is acceptable and the model is suitable for population-level prediction. The estimated residual standard deviation σ is 0.190, the bootstrap mean is 0.188, SD = 0.005, CV% = 2.893%, and the relative bias is -0.604%. Its 95% CI range is (0.178, 0.199), which is very narrow, suggesting that the model's intra-individual variation estimation has extremely high stability and reliability.

[0056] Table 1 Comparison of final model parameters and Bootstrap parameters

[0057] 3. Visual Predictive Check (VPC).

[0058] After the final model was validated using Bootstrap, it was further validated by performing 1000 simulations using the VPC method. The VPC diagnostic chart (IVAR vs. DV) is shown below. Figure 3 As shown, the 5th, 50th, and 95th percentiles of all observations fall within the 90% CI of the corresponding predicted values, indicating a high degree of agreement between the predicted and observed values ​​and suggesting that the model has good predictive performance.

[0059] 4. External validation of the perioperative PK model.

[0060] To comprehensively validate the applicability and predictive performance of the perioperative PK model constructed in this study in clinical practice, this study conducted external validation analyses from two dimensions: population predictive value (PRED) and individual predictive value (IPRED). Clinical data were collected from 32 hemophilia A patients who had previously undergone orthopedic surgery at Tongji Hospital affiliated with Tongji Medical College of Huazhong University of Science and Technology, the Third Affiliated Hospital of Guangzhou University of Chinese Medicine, and Shenzhen Second People's Hospital.

[0061] (1) Analysis of the accuracy of population predictive value (PRED). In the external validation of this study, population predictions (PREDs) of patients in the external validation data were directly calculated using fixed population parameters of the model. The median prediction error (MDPE), mean absolute prediction error (MAPE), and the proportion of predicted values ​​falling within ±20% and ±30% of the true value (F20 and F30) were used as evaluation indicators of the model's population prediction performance.

[0062] The MDPE of the Nanfang Model in this study was 23.9%, suggesting a certain degree of systematic overestimation at the population prediction level. This level of bias is similar to that of the Hazendonk model (24.7%) from the Netherlands, and slightly higher than the UNC model (14.2%) from Jing Zhu et al. in the United States. Nevertheless, the MAPE of the Nanfang Model was 28.7%, significantly lower than that of the Hazendonk model (56.7%) and the UNC model (41.5%), indicating that its overall absolute prediction error is smaller and its prediction accuracy is higher. In addition, the F20 and F30 of the Nanfang Model were 35.0% and 53.3%, respectively, meaning that more than half of the individual patient predictions were within ±30% of the true value, demonstrating good population prediction ability.

[0063] Table 2. Accuracy analysis of external validation population predictive values ​​(PRED) of the perioperative PK model.

[0064] (2) Analysis of the accuracy of Individual Predicted Values ​​(IPRED).

[0065] Based on Bayesian feedback, and considering different numbers of individual observations (0–2), the Median Individual Prediction Error (MDIPE), Mean Absolute Individual Prediction Error (MAIPE), and the proportion of prediction errors falling within ±20% (IF20) and ±30% (IF30) of the true value are calculated to comprehensively evaluate the model's individual prediction performance. MDIPE reflects the median trend of prediction bias, MAIPE measures overall prediction accuracy, and IF20 and IF30 describe the accuracy of predictions reaching the preset error range.

[0066] Without any observations (i.e., a priori prediction), the model's MDIPE was 22.5%, MAIPE was 27.2%, and IF20 and IF30 were 29.6% and 59.3%, respectively, indicating that the model's prediction bias was significant and its accuracy limited without individual information input. As the number of observations increased, individual prediction performance improved significantly: with one observation included, MDIPE decreased to 7.3%, MAIPE decreased to 21.9%, and IF20 and IF30 increased to 46.3% and 66.7%, respectively; with two observations included, MDIPE further decreased to 2.1%, MAIPE to 20.7%, and IF20 and IF30 increased to 48.1% and 85.2%, respectively. The overall trend indicates that with the inclusion of individual patient data, prediction bias gradually decreased, absolute error significantly decreased, and prediction accuracy continuously improved.

[0067] Table 3. Accuracy analysis of the Individual Predictive Values ​​(IPRED) of the perioperative PK model external validation.

[0068] 5. Pilot Study of Perioperative PK Model.

[0069] To preliminarily evaluate the feasibility and predictive performance of the perioperative PK model in clinical practice, this study conducted a small-sample exploratory trial before the formal prospective clinical trial.

[0070] In this exploratory trial, all three patients with hemophilia A received preoperative factor VIII supplementation, with a preoperative loading dose between 2500 and 3000 IU. Postoperative initial FVIII:C activity levels ranged from 122.0 to 128.9 IU / dL, all meeting the preoperative target levels. Specific surgical procedures included bilateral ankle arthroscopy (minor surgery), right hip replacement (major surgery), and debridement of osteomyelitis lesions in the proximal tibia combined with resection of inflammatory pseudotumor (major surgery), with operative times ranging from 96 to 175 minutes. Intraoperative blood loss varied considerably, with patients undergoing major surgery experiencing 1000 mL and 100 mL of blood loss, respectively, while patients undergoing minor surgery experienced very little blood loss (5 mL). Only one patient (hip replacement) required an additional 1000 IU of factor VIII due to significant intraoperative blood loss (1000 mL), while the other two patients did not receive additional factor doses. All patients received tranexamic acid (antifibrinolytic therapy) during the operation, and no red blood cell or plasma transfusions or NSAIDs were used.

[0071] The perioperative coagulation factor administration model for hemophilia A of this invention was applied to three patients according to the following steps, and the average factor usage of the three patients was finally calculated. Details are as follows: (1) Obtain the patient's pathological information, including information on the coagulation factor administration scenario and the covariates of the hemophilia A perioperative coagulation factor administration model of the present invention. Import the patient's covariate information and the data files of different administration scenarios to be simulated (including loading dose and administration after loading dose) into Phoenix NMLE software to perform Monte Carlo random simulation. Calculate the probability of achieving the perioperative coagulation factor concentration treatment target (e.g., the probability of activity above 80%) under different administration schemes. After multiple attempts, select the administration scheme with the highest probability of achieving the target as the final recommended administration scheme. (2) After administration of coagulation factor VIII according to the initial coagulation factor VIII administration regimen, measure the peak activity of coagulation factor VIII after the preoperative loading dose, assess whether the therapeutic window range has been reached, and closely monitor the patient's postoperative bleeding risk and wound recovery. (3) Input the peak activity of coagulation factor VIII after the preoperative loading dose, the dosing information in the initial coagulation factor VIII dosing regimen (including drug usage and dosage in the initial coagulation factor VIII dosing regimen), and the covariates of the hemophilia A perioperative coagulation factor dosing model of the present invention into the hemophilia A perioperative coagulation factor dosing model of the present invention. Estimate the individualized pharmacokinetic parameters of the patient by combining the maximum a posteriori Bayes method with prior information (including pharmacokinetic characteristic parameters of the target population, including: the mean of PPK parameters, the median of PPK parameters, the inter-individual variation of parameters, the intra-individual variation of parameters and / or the factors causing parameter variation) and observation data (including the activity monitored after perioperative coagulation factor VIII dosing) to obtain the individualized dosing regimen; (4) After administration of the individualized dosing regimen, perioperative coagulation factor VIII administration monitoring was performed to obtain the peak activity of coagulation factor VIII after the patient’s preoperative loading dose. Combined with the perioperative coagulation factor VIII activity treatment window, the individualized dosing regimen was given according to step (3). (5) Repeat steps (1)-(4) to ensure that the peak activity of coagulation factor VIII after the patient’s preoperative loading dose is within the therapeutic window at different stages of the perioperative period, thereby reducing the risk of perioperative bleeding.

[0072] Simultaneously, a preoperative pharmacokinetic (PK) guidance group and a weight guidance group were established. The preoperative PK guidance group's procedure involved administering a standard dose of FVIII (usually 25-50 IU / kg) to patients before surgery, and collecting blood samples at multiple time points (e.g., 0h, 1h, 4h, 8h, 24h, 48h) to measure FVIII activity (FVIII:C). The patients' FVIII concentration-time data were input into pharmacokinetic modeling software (e.g., Phoenix WinNonlin) to fit individual patient PK parameters, including volume of distribution (V), clearance (CL), and half-life (t½). Based on the hemostasis requirements at different perioperative stages (preoperative, intraoperative, and postoperative days 1–14), target FVIII:C levels were set (e.g., >100% intraoperatively, >80% on postoperative days 1–3, >50% on postoperative days 4–7, and >30% on postoperative days 8–14). Using the individual PK parameters mentioned above, combined with the target concentration curve and the planned dosing time points, the recommended dose and interval for each dosing are calculated through pharmacokinetic simulation (concentration-time curve under simulated dosing regimen), thus realizing the design of individualized dosing regimens. Weight guidance group: The dosage of the weight guidance group was calculated based on the patient's weight using the formula (shown below). Individual pharmacokinetic differences were not considered. The dosage was calculated based on the expected factor activity value, which carries the risk of overdose or insufficient activity.

[0073]

[0074] The three patients included in this exploratory trial recovered well postoperatively, with hospital stays ranging from 9 to 18 days. Sutures were removed on the 14th postoperative day in all cases. Regarding wound drainage, one patient did not have a drainage tube inserted, while the drainage volumes in the other two patients were 450 mL and 290 mL, respectively. All patients received tranexamic acid antifibrinolytic therapy postoperatively, with one patient also receiving NSAIDs. No patients underwent red blood cell or plasma transfusions. No postoperative complications such as massive hemorrhage, subcutaneous hematoma, infection, reoperation, or delayed wound healing occurred.

[0075] In this exploratory trial, the average factor usage guided by the perioperative PK model was 243.4 ± 62.9 IU / kg, significantly lower than that in the preoperative PK-guided group (432.8 ± 165.5 IU / kg) and the weight-guided group (498.0 ± 236.5 IU / kg). Regarding the proportion of coagulation factor costs, the proportion of coagulation factor-related costs in total hospitalization costs was 37.4% for patients receiving perioperative PK-guided medication, also lower than that in the weight-guided group (68.8%) and the preoperative PK-guided group (58.1%), suggesting that perioperative PK-guided medication has a potential advantage in reducing the economic burden of coagulation factors. The data for the preoperative PK-guided group and the weight-guided group were calculated based on a retrospective summary of previous clinical data, specifically by counting a number of patients who used weight-based recommended medication or preoperative PK-guided medication, and calculating the average factor usage for these patients. This study does not include the three patients included in this exploratory trial.

[0076] Regarding the variability of coagulation factor activity, the coefficients of variation for activity in two major surgeries guided by the perioperative PK model were 0.046 and 0.177 on postoperative days 1–3 (D1–D3), 0.074 and 0.109 on days 4–6 (D4–D6), and 0.109 on days 7–14 (D7–D6). 14 The coefficients of variation for activity were 0.051 and 0.156, respectively, both lower than the activity variation levels of the preoperative PK-guided group and the weight-guided group at the corresponding time points. The activity variation coefficient of one patient undergoing minor surgery was 0.257 on postoperative days 1–5 (D1–D5), which was similar to that of the preoperative PK-guided group (0.222) and the weight-guided group (0.236).

[0077] Regarding the achievement of activity targets, the failure rate of activity values ​​in the preoperative PK-guided group was 64.4%, and in the weight-guided group it was 70.1%. However, the failure rates of activity values ​​in the three patients guided by the perioperative PK model were 20.0%, 33.3%, and 30.4%, respectively, with an average of 27.9%, which was lower than that in the preoperative PK-guided group and the weight-guided group. This suggests that perioperative PK model guidance can help improve the achievement rate of activity targets and further optimize the management of coagulation factors.

[0078] Table 4 Comparative Analysis of Coagulation Factor Usage, Activity Variability, and Activity Target Achievement Rate under Three Dosing Guidance Strategies

[0079] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0080] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A perioperative coagulation factor administration model for hemophilia A, characterized in that, The perioperative coagulation factor administration model for hemophilia A was constructed using a nonlinear mixed-effects modeling method based on a two-compartment model structure. The covariates of the perioperative coagulation factor administration model for hemophilia A include blood type, weight, age, use of perioperative antifibrinolytic drugs, and use of perioperative NSAID analgesics.

2. The perioperative coagulation factor administration model for hemophilia A according to claim 1, characterized in that, The perioperative coagulation factor administration model for hemophilia A is shown below: V 2873.47 (0.28 ) 4093.37 192.21 Baseline value 1.09 Where V: central compartment distribution volume, unit: mL, representing the initial distribution volume of the drug; I Antifibrinolytic drug Perioperative antifibrinolytic drug use: A value of 1 indicates that antifibrinolytic drugs were used during the perioperative period, and a value of 0 indicates that antifibrinolytic drugs were not used during the perioperative period. V2: Peripheral compartment distribution volume, unit: mL, representing the volume of drug distributed in peripheral tissues; CL: Central compartment clearance rate, unit: mL / h, represents the volume of drug cleared from the central compartment per unit time, which is affected by individual characteristics and covariates; I Blood Type The patient's blood type: a value of 1 indicates that the patient's blood type is O, and a value of 0 indicates that the patient's blood type is not O. CL2: Distribution clearance rate of peripheral compartments, unit: mL / h, representing the exchange rate between the central compartment and the peripheral compartment; I Antiinflammatory drug The value of 1 indicates that NSAID analgesics were used during the perioperative period, and the value of 0 indicates that NSAID analgesics were not used during the perioperative period. Baseline value: Peak activity of coagulation factor VIII after preoperative loading dose, representing the initial concentration before surgery; η represents the inter-individual variation corresponding to each parameter.

3. The application of the perioperative coagulation factor administration model for hemophilia A according to any one of claims 1-2 in drug administration guidance, characterized in that, The application includes the following steps: (1) Obtain the patient’s perioperative information, which includes coagulation factor administration scenario information and covariates of the hemophilia A perioperative coagulation factor administration model according to any one of claims 1-2, and obtain the initial coagulation factor VIII administration regimen under the administration scenario through Monte Carlo random simulation calculation; (2) After administration of the initial coagulation factor VIII according to the prescribed administration regimen, the peak activity of coagulation factor VIII after the preoperative loading dose is measured to assess whether the therapeutic window range has been reached. (3) Input the peak activity of coagulation factor VIII after the preoperative loading dose, the dosing information in the initial coagulation factor VIII dosing regimen, and the covariates of the perioperative coagulation factor dosing model of hemophilia A according to any one of claims 1-2 into the perioperative coagulation factor dosing model of hemophilia A, and estimate the individualized pharmacokinetic parameters of the patient by combining the maximum a posteriori Bayes method with prior information and observation data to obtain the individualized dosing regimen; (4) After administration of the individualized dosing regimen, perioperative coagulation factor VIII administration monitoring is performed to obtain the peak activity of coagulation factor VIII after the patient’s preoperative loading dose. Combined with the perioperative coagulation factor VIII activity treatment window, the individualized dosing regimen is given according to step (3). (5) Repeat steps (1)-(4) to ensure that the peak activity of coagulation factor VIII after the patient’s preoperative loading dose is within the therapeutic window at different stages of the perioperative period, thereby reducing the risk of perioperative bleeding.

4. The application according to claim 3, characterized in that, In step (3) of the application, the drug administration information in the initial coagulation factor VIII administration regimen is the drug usage and drug dosage in the initial coagulation factor VIII administration regimen.

5. A system for providing a coagulation factor administration regimen, characterized in that, include: Data storage module for storing patient information and the perioperative coagulation factor administration model for hemophilia A as described in any one of claims 1-2; A data analysis module is used to obtain a personalized dosing regimen according to steps (1)-(4) in any one of claims 3-4; and The data display module is used to show patients' individualized medication regimens.

6. The system according to claim 5, characterized in that, The information includes: blood type, weight, age, use of antifibrinolytic drugs, and use of NSAID analgesics.

7. The system according to claim 5, characterized in that, The patient is a hemophilia A patient.

8. The system according to claim 7, characterized in that, The patient was in the perioperative period.

Images