基于生理药物代谢动力学模型预测氯氮平联合用药的药物相互作用
1.
2.
3.
Prediction of drug-drug interactions in clozapine combination therapy based on physiologically based pharmacokinetic model
1.
2.
3.
通讯作者: 禹顺英,电子信箱:yushunying@smhc.org.cn。
编委: 崔黎明
收稿日期: 2024-05-22 接受日期: 2024-06-24 网络出版日期: 2024-11-28
基金资助: |
|
Corresponding authors: YU Shunying, E-mail:yushunying@smhc.org.cn.
Received: 2024-05-22 Accepted: 2024-06-24 Online: 2024-11-28
目的·以氯氮平-氟伏沙明合用为例,通过构建针对中国群体的生理药物代谢动力学(physiologically based pharmacokinetic,PBPK)模型,预测氯氮平联合用药的药物相互作用(drug-drug interaction,DDI)并对氯氮平进行剂量优化。方法·通过文献及药理学相关数据库获取氯氮平及氟伏沙明的基本理化性质参数,药物吸收、分布、代谢及排泄(absorption,distribution, metabolism and excretion,ADME)相关参数及中国群体的生理解剖相关参数,利用PK-Sim®软件构建2种药物的PBPK模型。以平均百分比误差(mean percentage error,MPE)和平均绝对百分比误差(mean absolute percentage error,MAPE),或者预测药时曲线下面积(area under the curve,AUC)或峰浓度(peak concentration,Cmax)与实测AUC或Cmax的比值为判断指标,并通过真实世界血药浓度数据进行模型验证。在此基础上结合氟伏沙明对氯氮平的抑制作用参数构建氯氮平-氟伏沙明联合用药的PBPK模型,预测氯氮平的药物代谢动力学变化。以药时曲线下面积比值(area under the curve ratio,AUCR)或峰浓度比值(peak concentration ratio,CmaxR)的90%置信区间为评价指标判断是否存在临床显著的DDI(无效应边界为80%~125%)。根据PBPK模型量化氯氮平-氟伏沙明联合用药后氯氮平的药物代谢动力学变化,并制定氯氮平的剂量优化方案。结果·构建的氯氮平、氟伏沙明模型验证的MPE绝对值≤10%且MAPE<25%,说明预测的药时曲线是准确的。氯氮平-氟伏沙明合用的PBPK模型的AUC预测值与实测值的比值在1.25以内,可准确地预测药物代谢动力学参数。氯氮平-氟伏沙明联用模型的预测结果提示,氯氮平-氟伏沙明联合用药的AUCR和CmaxR的90%置信区间均不完全位于无效应边界内,说明两药合用会发生临床显著性的DDI。此外,PBPK模型的剂量优化结果提示:受试者联合服用氯氮平及氟伏沙明时,氯氮平的剂量减少至原本剂量的50%,可使氯氮平的暴露水平与单药治疗时保持一致。结论·研究建立的PBPK模型可以较好模拟联合用药对氯氮平药物代谢动力学的影响,对于预测药物可能的相互作用及剂量优化方案有参考意义。如果治疗过程中需要合用氯氮平和氟伏沙明,须警惕临床显著的DDI,并应优化氯氮平的剂量。
关键词:
Objective ·To develop physiologically based pharmacokinetic (PBPK) models specifically designed for the Chinese population by utilizing the combination of clozapine and fluvoxamine as a case, and predict the drug-drug interaction (DDI) associated with the combination medication of clozapine, ultimately optimizing the dosage of clozapine. Methods ·By obtaining the physicochemical parameters, absorption, distribution, metabolism, excretion (ADME)-related parameters, and physiologically relevant parameters of the Chinese population through literature and pharmacology-related databases, PBPK models for the clozapine and fluvoxamine were constructed by using PK-Sim® software. The models′ accuracy was evaluated by comparing predicted values of the area under the curve (AUC) and peak concentration (Cmax) to observed data, using the mean percentage error (MPE) and mean absolute percentage error (MAPE) as evaluation indicators. The models were validated against real-world plasma drug concentration data. Additionally, combining the inhibitory effect of fluvoxamine on clozapine, models for the combination therapy of clozapine and fluvoxamine were developed to predict the pharmacokinetic changes of clozapine. The presence of clinically significant DDI was determined by using the 90% confidence interval of the AUC ratio (AUCR) or Cmax ratio (CmaxR) as evaluation metrics, with a non-effect boundary set at 80%‒125%. The pharmacokinetic changes of clozapine upon co-administration with fluvoxamine based on PBPK models were quantified, and a dosage optimization for clozapine was developed. Results ·The constructed model of clozapine and fluvoxamine was considered accurate if the absolute value of the MPE was ≤10% and the MAPE was <25% during validation, indicating that the predicted concentration-time curves were accurate. The PBPK model for the co-administration of clozapine and fluvoxamine was able to accurately predict pharmacokinetic parameters if the ratio of predicted AUC to observed AUC was within 1.25. The prediction of PBPK model for the co-administration showed that the 90% confidence intervals for AUCR and CmaxR of the combination therapy of clozapine and fluvoxamine were not entirely within the ineffective effect boundary, indicating a clinically significant DDI when these two drugs were used concomitantly. Moreover, the dose optimization according to the PBPK models indicated that when subjects were co-administered with clozapine and fluvoxamine, reducing the dose of clozapine to 50% of the original dose could maintain the exposure levels of clozapine consistent with monotherapy. Conclusion ·The established PBPK model can effectively simulate the impact of combination therapy on pharmacokinetic changes of clozapine, providing valuable insights for predicting potential DDI and optimizing dosage regimens. If clozapine needs to be co-administered with fluvoxamine during the treatment, clinicians should remain vigilant for clinically significant DDI and contemplate optimizing the dosage of clozapine accordingly.
Keywords:
本文引用格式
牟凡, 黄志伟, 程渝, 赵雪, 李华芳, 禹顺英.
MOU Fan, HUANG Zhiwei, CHENG Yu, ZHAO Xue, LI Huafang, YU Shunying.
氯氮平是难治性精神分裂症优先考虑的药物。氯氮平与其他药物联合使用时,可能会影响氯氮平的血药浓度,甚至会导致癫痫发作、代谢综合征等药物不良反应(adverse drug reaction,ADR)的发生[1-2]。氟伏沙明是细胞色素P450成员1A2(cytochrome P450 1A2,CYP1A2)的强抑制剂,与氯氮平合用会发生药物-药物相互作用(drug-drug interaction,DDI),从而增加药物不良反应的发生风险[3-4]。美国食品药品监督管理局(Food and Drug Administration,FDA)的药物标签建议合用氟伏沙明时减少氯氮平的剂量[5],该结论是针对高加索人群得出的,是否适用于中国群体尚无明确的证据[6-7]。生理药物代谢动力学(physiologically based pharmacokinetic,PBPK)模型是药物代谢动力学研究的重要工具,能用于研究合并用药、种族等影响因素对药物体内给过程的影响[8-10]。若能针对中国群体构建PBPK模型,对合用氟伏沙明后氯氮平的血药浓度进行预测,并以此为依据提出血药浓度剂量优化建议,可降低氯氮平的治疗风险[11-13]。
综上,本研究希望通过针对中国群体的氯氮平-氟伏沙明合用的PBPK模型,量化合并用药对氯氮平血药浓度的影响,以期在临床存在合用氟伏沙明的需求时,为调整氯氮平剂量提供参考。
1 材料与方法
1.1 PBPK模型的建立与验证
1.1.1 PBPK模型的软件平台
PK-Sim®软件(10.0版)用于PBPK模型的建立以及参数的优化,由德国拜尔公司研发。WebPlotDigitizer软件(4.5版)用于提取临床数据中药时曲线数据的信息,由美国圣母大学开发完成。OriginPro®(9.5.5版)对输出的结果进行数据分析以及图形编辑,由美国OriginLab公司研发。
1.1.2 建模步骤与数据来源
PBPK模型的建模数据主要来自文献及数据库,PK-Sim®软件将模型分为群体、药物、给药方案及剂型共4个模块,它们共同模拟了药物在体内的过程。
其中,WM是数据的加权平均值,nj 指每个数据的样本量,xj 指每个数据中的酶的平均丰度。
表1 中国人群各CYP450亚型的酶含量
Tab 1
CYP450 | Sample 1[14] | Sample 2[15] | Sample 3[16] | WM |
---|---|---|---|---|
CYP1A2/(pmol·mg-1) | 42.48 | 42.00 | 42.30 | 42.31 |
CYP2A6/(pmol·mg-1) | 15.63 | - | - | 15.63 |
CYP2B6/(pmol·mg-1) | 4.62 | - | - | 4.62 |
CYP2C9/(pmol·mg-1) | 98.60 | - | 87.20 | 87.19 |
CYP2C19/(pmol·mg-1) | 8.45 | 60.00 | 8.10 | 8.25 |
CYP2D6/(pmol·mg-1) | 20.50 | - | - | 20.50 |
CYP2E1/(pmol·mg-1) | 102.04 | - | - | 102.04 |
CYP3A4/(pmol·mg-1) | 49.34 | 120.00 | 93.00/70.30 | 49.34 |
CYP3A5/(pmol·mg-1) | 42.45 | - | 145.40/82.10 | 42.45 |
从相应的临床试验中获取年龄、性别、身高及体质量等人口学特征的信息以构建虚拟群体;若缺乏相关信息则从美国国家健康与营养调查(National Health and Nutrition Examination Survey,NHANES)数据库中获取。
(2)药物模块及给药方案模块。在药物数据库(DrugBank)等药理学数据库、FDA以及文献中收集药物的基本理化性质参数及药物吸收、分布、代谢及排泄(absorption,distribution,metabolism and excretion,ADME)相关参数来建立药物模块。经文献检索和各药理学数据库检索收集的氯氮平及氟伏沙明的基本理化性质参数与ADME相关参数详见表2。氯氮平的给药方案有2种,分别是:150 mg/次,2次/d;100 mg/次,2次/d。
表2 氯氮平和氟伏沙明的基本理化性质及ADME相关参数
Tab 2
Parameter | Clozapine | Source | Fluvoxamine | Source |
---|---|---|---|---|
LogP | 3.23 | DrugBank | 3.68 | Fitted |
fu | 0.02 | FDA, fitted | 0.20 | DrugBank, fitted |
Molecular weight/(g·mol-1) | 326.83 | DrugBank | 318.34 | DrugBank |
pKa | 7.50 | DrugBank | 9.40 | Fitted |
Solubility/(mg·mL-1) | 0.19 | ALOGPS | 0.07 | Fitted |
CLh/(mL·h-1·kg-1) | 2.28 | Fitted | 1.21 | Fitted |
CLr/(mL·h-1·kg-1) | 0.01 | Fitted | 0.02 | FDA, fitted |
(3)剂型模块。当给药途径为口服时需要建立剂型模块以模拟药物在胃肠道内的溶出过程,该过程一般通过威布尔模型描述[17]。
其中,a为尺度参数,b为形状参数,m为t时药物溶解的百分比,Tlag为药物溶出的迟滞时间(单位h)。
1.1.3 模型验证
(1)模型验证的评价指标。PBPK模型可以从2个方面进行验证。一是以平均百分比误差(mean percentage error,MPE)和平均绝对百分比误差(mean absolute percentage error,MAPE)为评价依据,当MPE绝对值≤10%且MAPE<25%时模型预测的药时曲线是准确的(公式3、4)[18]。二是以预测药时曲线下面积(area under the curve,AUC)或峰浓度(peak concentration,Cmax)与实测AUC或Cmax的比值为判断指标,当预测值与实测值的比值在1.25倍内可认为预测药物代谢动力学参数的准确度较高。
其中,n为样本量,
1.2 氯氮平-氟伏沙明联合用药的PBPK模型
1.2.1 DDI模型的构建与验证
1.2.2 基于PBPK模型的DDI预测
(2)氯氮平的剂量优化。氯氮平与氟伏沙明合用后,以氯氮平在稳态情况下的Cmax以及Ctrough为依据调整氯氮平剂量。当两药合用后氯氮平的Cmax及Ctrough均与单用氯氮平保持一致时,此时氯氮平的剂量为优化剂量。
2 结果
2.1 单药的PBPK模型及验证结果
2.1.1 氯氮平的PBPK模型及验证结果
研究模拟多次口服氯氮平(100 mg/次,2次/d)的PBPK模型并进行验证。内部验证的结果显示,氯氮平的血药浓度达到稳态后AUC均值为4 388.90 ng·h/mL,Cmax均值为516.89 ng/mL,Ctrough均值为129.57 ng/mL(图1)。内部验证的结果表明,模型预测的MPE值为-7.66%,MAPE为9.71%,表明PBPK模型预测氯氮平的药物代谢动力学参数准确度较高。
图1
图1
健康受试者多次口服100 mg氯氮平后氯氮平的药时曲线及残差图
Note: A. Plasma concentration-time curve profiles of clozapine after the administration of multiple oral 100 mg tablets in healthy adults. B. Residue plot for the model prediction of clozapine plasma concentrations in healthy adults.
Fig 1
Plasma concentration-time curve profiles and residue plot of clozapine after multiple oral doses of 100 mg in healthy adults
外部验证数据来自上海交通大学医学院附属精神卫生中心病历系统,用于进一步评价PBPK模型的预测能力。模型预测了2个给药方案(150 mg/次,2次/d;100 mg/次,2次/d)。前一个给药方案的AUC均值为4 402.18 ng·h/mL,Cmax均值为817.08 ng/mL,Ctrough均值是271.27 ng/mL;后一个给药方案的AUC均值为3 597.09 ng·h/mL,Cmax均值为522.93 ng/mL,Ctrough均值为150.34 ng/mL(图2)。外部验证数据中符合2个给药方案的血药浓度数据共32个,74.36%的数据位于预测值的95%置信区间内。
图2
图2
患者多次口服氯氮平片后的药时曲线
Note: A. Plasma concentration-time curve profiles of clozapine after the administration of multiple 150 mg oral tablets in patients. B. Plasma concentration-time curve profiles of clozapine after the administration of multiple 100 mg oral tablets in patients.
Fig 2
Plasma concentration-time curve profiles of clozapine after the administration of multiple oral tablets of clozapine in patients
2.1.2 氟伏沙明的PBPK模型与验证结果
研究模拟了单次口服50 mg氟伏沙明的PBPK模型,预测的AUC均值为353.89 ng·h/mL,Cmax均值为15.92 ng/mL。模型内部验证结果表明,模型预测的MPE值为-6.99%,MAPE为11.81%,说明PBPK模型预测氟伏沙明的药物代谢动力学参数的准确度高(图3)。
图3
图3
健康受试者单次50 mg口服氟伏沙明后的药时曲线及残差
Note: A. Plasma concentration-time curve profiles of fluvoxamine after the administration of a single 50 mg dose of fluvoxamine in healthy adults. B. Residue plot for the model prediction of fluvoxamine plasma concentrations in healthy adults.
Fig 3
Plasma concentration-time curve profiles and residue plot of fluvoxamine after the administration of a single 50 mg dose of fluvoxamine in healthy adults
2.2 氯氮平-氟伏沙明联合用药
2.2.1 氯氮平-氟伏沙明联合用药的PBPK模型与验证
氟伏沙明-氯氮平的联合用药模型预测了氯氮平血药浓度达到稳态后的暴露水平,2个给药方案的AUC均值分别为10 380.99 ng·h/mL和10 288.17 ng·h/mL,Cmax均值分别为1 049.12 ng/mL和1 663.57 ng/mL,Ctrough均值分别为712.49 ng/mL和1 460.93 ng/mL。PBPK模型的模拟结果表明合用氟伏沙明的AUCR为2.11,既往临床试验的AUCR为2.3[23],即Rpre/obs为0.92(<1.25),说明DDI的预测效果佳。另一项临床研究结果[3]表明氯氮平合用氟伏沙明后,氯氮平的血药浓度/剂量比值(plasma concentration/dose,C/D)较单药使用时增加了117.9%,模型预测结果显示氯氮平的C/D为111.4%,也侧面验证了模型预测的准确性。
2.2.2 基于PBPK模型的DDI预测结果及剂量优化
表3 氯氮平-氟伏沙明PBPK模型药物代谢动力学参数比较
Tab 3
Drug administration | Pharmacokinetic parameter | Combination therapy/monotherapy | No effect boundary | |
---|---|---|---|---|
mean/% | 90%CI | |||
Administration 1 | Cmax/(ng·mL-1) | 180 | 111%‒223% | 80%‒125% |
AUC/(ng·h·mL-1) | 197 | 105%‒204% | ||
Administration 2 | Cmax/(ng·mL-1) | 175 | 110%‒240% | |
AUC/(ng·h·mL-1) | 189 | 100%‒277% |
图4
图4
氯氮平-氟伏沙明合用前后氯氮平的药时曲线的比较
Note: A. Plasma concentration-time curve profiles of clozapine (100 mg) before and after combination with fluvoxamine (50 mg). B. Plasma concentration-time curve profiles of clozapine (150 mg) before and after combination with fluvoxamine (50 mg). bid—twice a day; qd—once a day.
Fig 4
Comparison of plasma concentration-time curve profiles of clozapine before and after combination with fluvoxamine
(2)剂量优化。以受试者口服氯氮平100 mg/次、2次/d及氟伏沙明50 mg/次、1次/d为例,经模型拟合制定的剂量调整建议是:受试者在加用50 mg氟伏沙明后氯氮平片的剂量从100 mg(2次/d)可以减为50 mg(2次/d)。在这种情况下,加用氟伏沙明后氯氮平的暴露水平与单药使用氯氮平时保持一致(图5)。
图5
图5
剂量优化前后氯氮平的药时曲线比较
Fig 5
Comparison of plasma concentration-time curve profiles of clozapine before and after dose optimization
3 讨论
研究构建了针对中国群体的氯氮平、氟伏沙明及两药合用的PBPK模型,建立的模型均能较为准确地预测药物在体内的药物代谢动力学过程。在模型验证中,研究通过上海交通大学医学院附属精神卫生中心病历系统,选取了2018—2019年期间与模型的虚拟群体相匹配的患者,提取了氯氮平的血药浓度数据进行外部验证,更为有力地证明了PBPK模型可以较好地预测中国群体的氯氮平的体内过程。在构建的模型基础上,研究对氯氮平-氟伏沙明合并用药进行了DDI预测,结果提示两药合用存在显著的DDI。因此若需合用氟伏沙明,应减少氯氮平的剂量以提高用药安全性。FDA的药物标签提出氯氮平与氟伏沙明合用时应减至原始剂量的1/3[5]。需要注意的是,该结论主要基于高加索人群的数据,而氯氮平是一个在种族间存在显著代谢差异的典型药物。亚洲人对于氯氮平的代谢能力更弱,在合并氟伏沙明时更应警惕血药浓度的变化[32-33]。研究结果提示,在中国群体中,若将氯氮平的剂量调整为原来的50%,可以使加用氟伏沙明后氯氮平的暴露水平与单药治疗时一致。
此外,本研究也具有一些局限性:首先,氯氮平是多代谢途径以及多代谢产物的药物,代谢过程十分复杂,是研究氯氮平体外代谢的难点之一。因此研究构建氯氮平的PBPK模型只对氯氮平药物代谢动力学进行了预测。未来需要通过体外试验或者大量的临床数据拟合不同的代谢酶催化生成各代谢产物的过程。其次,虽然PBPK模型的内部验证与外部验证结果均提示模型预测性能较好,但是作为一个预测模型的研究,还需要更多的临床试验数据来对预测结果进行验证和修正。
综上,本研究基于中国群体,从药物代谢动力学角度提出了氯氮平-氟伏沙明联合用药时的剂量优化方法,为临床试验或者临床给药方案的制定提供了参考。
作者贡献声明
禹顺英设计并指导整个课题研究,牟凡参与研究设计并完成数据收集、模型构建及验证和论文撰写,黄志伟参与模型构建及验证的指导,程渝、赵雪参与数据收集与整理,黄志伟、李华芳和禹顺英参与论文修改。所有作者均阅读并同意了最终稿件的提交。
AUTHOR's CONTRIBUTIONS
YU Shunying designed and supervised the entire research project; MOU Fan participated in research design, and completed data collection, model construction, validation, and paper writing; HUANG Zhiwei participated in guiding model construction and validation; CHENG Yu and ZHAO Xue participated in data collection and organization; HUANG Zhiwei, LI Huafang and YU Shunying participated in paper revisions. All the authors have read the last version of the manuscript and consented to submission.
利益冲突声明
所有作者声明不存在利益冲突。
COMPETING INTERESTS
All authors disclosed no relevant conflict of interests.
参考文献
/
〈 | 〉 |