摘要:目的 本研究旨在對前期在皮膚軟組織和骨關(guān)節(jié)感染的兒童患者中建立的萬古霉素群體藥物動力學(xué)模型進(jìn)行外部驗證,從而判斷其是否可以外推至其他患者人群,用于指導(dǎo)萬古霉素在兒童患者中的個體化應(yīng)用。方法 從醫(yī)院信息系統(tǒng)中回顧性提取2021年6月—2022年12月入住我院兒童患者的年齡、性別、體重、實驗室檢查和萬古霉素血藥濃度等基本信息。將前期建立的兒童患者群體藥物動力學(xué)模型在Phoenix軟件中重建,通過貝葉斯法預(yù)測得到每個患者萬古霉素的個體預(yù)測值,計算個體預(yù)測值與觀測值的相對預(yù)測誤差和均方根誤差(百分比)等統(tǒng)計量和作圖分析,對已建立的萬古霉素群體藥代動力學(xué)模型進(jìn)行外部驗證。結(jié)果 共342例兒童患者的399個萬古霉素穩(wěn)態(tài)血藥濃度數(shù)據(jù)用于最終的外部驗證。除新生兒患者外,嬰幼兒、兒童及青少年患者萬古霉素血藥濃度觀測值及預(yù)測值計算獲得平均預(yù)測誤差和預(yù)測誤差中位數(shù)均在±20%之內(nèi),均方根誤差(百分比)在30%之內(nèi)。根據(jù)感染類型分層,除新生兒患者外,呼吸系統(tǒng)感染、中樞神經(jīng)系統(tǒng)感染及其他感染兒童患者萬古霉素血藥濃度觀測值及預(yù)測值計算獲得平均預(yù)測誤差和預(yù)測誤差中位數(shù)均在±20%之內(nèi),均方根誤差(百分比)在30%之內(nèi)(呼吸系統(tǒng)感染患者為31.1%)。結(jié)論 前期建立的兒童患者萬古霉素群體藥物動力學(xué)模型可以外推至嬰幼兒、兒童及青少年患者中用于模型引導(dǎo)的精準(zhǔn)用藥,從而指導(dǎo)萬古霉素在兒童患者中的個體化應(yīng)用。
關(guān)鍵詞:群體藥物動力學(xué)模型;外推;萬古霉素;兒童;模型引導(dǎo)的精準(zhǔn)用藥
中圖分類號:R978 文獻(xiàn)標(biāo)志碼:A
Extrapolating the vancomycin population pharmacokinetic model for model-informed precision dosing in pediatric patients supported by external validation
Lju Meng, Zhou Yuxue, Wang Danshu, and Zhang Shengnan
(Department of Pharmacy, Children's Hospital Affiliated to Zhengzhou University, Zhengzhou 450000)
Abstract" " Objective" " "This study aimed to evaluate the predictive performance of the previously estimated vancomycin population pharmacokinetic (PopPK) model using data from pediatric populations with skin, soft tissue, and bone and joint infections, thus determining whether it can be extrapolated to other patient populations and guiding the individualized application of vancomycin in children. Methods" " The basic characteristics of pediatric patients, including age, gender, body weight, laboratory examination, and their blood concentrations of vancomycin from June 2021 to December 2022, were retrospectively extracted from the hospital information system. The pre-established population pharmacokinetic model of the pediatric patient group was reconstructed in the Phoenix software, and the individual predictive value of vancomycin for each patient was obtained through Bayes forecast, including the mean relative prediction error [MPE (%)], median relative prediction error [MDPE (%)], and root mean squared error (in percentage) (RMSE) for individual predictions, and external validation was carried out on the established pharmacokinetics model. Results" " A total of 399 vancomycin concentrations at steady state from 342 pediatric patients were included in the final external validation dataset. With the exception of newborns, based on observed and predicted values, the MPE (%) and MDPE (%) in infants, children and adolescents were within ±20%, and the RMSE was within 30%. Based on the type of infection, the MPE (%) and MDPE (%) were within ±20% and the RMSE was within 30% (31.1% in patients with respiratory infection) in pediatric patients with respiratory infection, central nervous system infection and other infections, excluding neonatal patients. Conclusion" " The previously established vancomycin PopPK model could be used in clinical practice in infants, children and adolescents to optimize vancomycin precise dosing, thereby guiding the individualized application of vancomycin in children.
Key words" " Population pharmacokinetic model; Extrapolation; Vancomycin; Pediatric patients; Model-informed precision dosing
萬古霉素為糖肽類抗生素,通過抑制細(xì)菌細(xì)胞壁的合成發(fā)揮抗菌作用,是臨床革蘭陽性菌感染尤其是耐藥金黃色葡萄球菌感染(methicillin-resistant Staphylococcus aureus, MRSA)的一線用藥[1]。藥動學(xué)/藥效學(xué)研究表明,萬古霉素24 h藥時曲線下面積與最小抑菌濃度的比值(area under the curve over 24 hours/minimum inhibitory concentration, VAN-AUC24/MIC)、穩(wěn)態(tài)谷濃度與其臨床療效和安全性最相關(guān)[2-3]。因此,萬古霉素最新治療藥物監(jiān)測指南推薦臨床采用模型引導(dǎo)的精準(zhǔn)用藥指導(dǎo)萬古霉素個體化應(yīng)用,使VAN-AUC24維持在400~600/650 mg/(h.L)(假設(shè)MIC為1 mg/L)
以及谷濃度為5~15 mg/L,從而產(chǎn)生較好的臨床療效并且減少腎毒性的發(fā)生[4-6]。
AUC通常采用經(jīng)典的梯形法計算獲得,但由于該方法需要在同一個給藥間隔采集多個血藥濃度數(shù)據(jù)點(diǎn),對于兒童患者來說臨床可操作性較低。近年來,通過群體藥物動力學(xué)(population pharmacokinetic, PopPK)模型結(jié)合臨床常規(guī)血藥濃度監(jiān)測數(shù)值,采用貝葉斯法估算個體藥動學(xué)參數(shù),從而計算獲得AUC,已經(jīng)越來越多應(yīng)用于臨床[7]。選擇合適的模型是準(zhǔn)確預(yù)測個體藥動學(xué)參數(shù)的關(guān)鍵。因此,在外推使用該模型進(jìn)行模型引導(dǎo)的精準(zhǔn)用藥之前對其進(jìn)行綜合全面的外部驗證顯得尤為重要[8-9]。
本研究團(tuán)隊前期通過收集25例兒童皮膚軟組織和骨關(guān)節(jié)感染的兒童患者的75份剩余血樣,建立萬古霉素群體藥物動力學(xué)模型。該模型為一室模型,清除率(clearance, CL)和表觀分布容積(volume, V)的群體典型值分別為0.14 L/(h.kg)和0.5 L/kg[10]。該模型群體典型值與不同群體兒童患者中萬古霉素群體典型值的中位數(shù)(CL, 0.103 L/(h.kg); V, 0.64 L/kg)較為接近[10-11]。因此,該模型是否可以外推至其他兒童患者人群中亟待進(jìn)行驗證。本研究通過回顧性收集臨床萬古霉素血藥濃度監(jiān)測數(shù)據(jù),對該模型進(jìn)行外部驗證,從而判斷其是否可以外推至其他兒童患者人群進(jìn)行模型引導(dǎo)的精準(zhǔn)用藥。
1 材料與方法
1.1 臨床數(shù)據(jù)收集
回顧性收集鄭州大學(xué)附屬兒童醫(yī)院2021年6月至2022年12月使用萬古霉素患兒的基本信息,包括年齡、性別、體重、萬古霉素使用劑量、萬古霉素血藥濃度數(shù)據(jù)、血清肌酐值。納入標(biāo)準(zhǔn):①萬古霉素使用≥3 d;②至少有1個萬古霉素血藥濃度觀測數(shù)值。排除標(biāo)準(zhǔn):①透析患者;②使用體外膜肺患者;③嚴(yán)重腎功能不全患者;④病例信息不全者;⑤院外有萬古霉素治療歷史。根據(jù)《國家抗微生物治療指南(第3版)》對患兒的感染類型進(jìn)行分類,如果涉及到多部位感染,以主要臨床診斷統(tǒng)計。萬古霉素血藥濃度采用酶放大免疫法(enzyme-multiplied immunoassay technique, EMIT)檢測,檢測儀器為SYVA Viva-E(西門子有限公司),線性范圍為2~
50 mg/L。每日進(jìn)行內(nèi)部質(zhì)控,主要為11.2和29.8 mg/L
兩個水平。本研究經(jīng)過鄭州大學(xué)附屬兒童醫(yī)院醫(yī)學(xué)倫理委員會批準(zhǔn)(批件號為2023-K-154)。
1.2 模型的重建及個體預(yù)測值的計算
通過Phoenix NLME (version 8.3.5, Certara)軟件對前期建立的模型進(jìn)行重建,模型結(jié)果及參數(shù)如下。采用貝葉斯反饋方法,求出與觀測值同樣時間點(diǎn)的個體濃度預(yù)測值,與實測值比較。采用GraphPad Prism 8(GraphPad Software, San Diego, CA, USA)軟件分別繪制PE(%)的箱線圖和觀測值與預(yù)測值的Bland-Altman圖。
(1)
(2)
式中BW為體重。
1.3 模型的外部驗證
通過等式3~5計算預(yù)測誤差[prediction error percentage, PE(%)]、平均預(yù)測誤差[mean relative prediction error, MPE(%)]以及預(yù)測誤差中位數(shù)[median relative prediction error, MDPE(%)]評價模型預(yù)測的偏倚,即模型預(yù)測的準(zhǔn)確度。通過等式6計算均方根誤差(the root mean squared error, RMSE)評價模型預(yù)測的精確度。MPE(%)和MDPE(%)在±20%之內(nèi),RMSE在30%之內(nèi)表明該模型的預(yù)測性能較好,可以在臨床用于模型引導(dǎo)的精準(zhǔn)用藥[12-13]。
(3)
(4)
MDPE (%) = median PEi(%)" " " " " " " " " " " " " " " " " " " " " (5)
(6)
N是觀察值的總數(shù),obsi和predi分別指第i個個體的觀測值和個體預(yù)測值。
2 研究結(jié)果
根據(jù)納入和排除標(biāo)準(zhǔn),共342例兒童患者的399個萬古霉素穩(wěn)態(tài)血藥濃度數(shù)據(jù)用于最終的外部驗證,其基本情況見表1。45.9%(183例)患兒為呼吸系統(tǒng)感染,其次為中樞神經(jīng)系統(tǒng)感染(18.8%,75例)。83.3%(285例)患兒僅有1個萬古霉素血藥濃度觀測數(shù)據(jù)。除5個萬古霉素血藥濃度觀測數(shù)據(jù)為非穩(wěn)態(tài)谷濃度數(shù)值外,其余均為谷濃度觀測數(shù)值。與建模的數(shù)據(jù)相比,納入外部驗證的數(shù)據(jù)年齡、體重、肌酐范圍更寬。
外部驗證的結(jié)果通過作圖和計算觀測值和個體預(yù)測值之間的MPE(%)和MDPE(%)形式呈現(xiàn)。PE(%)按照年齡和感染類型分層觀測值的箱線圖見圖1,統(tǒng)計量計算結(jié)果見表2。結(jié)果顯示,除新生兒外,嬰幼兒、兒童及青少年患者萬古霉素血藥濃度觀測值及預(yù)測值計算獲得MPE(%)和MDPE(%)均在±20%之內(nèi),RMSE在30%之內(nèi)。根據(jù)感染類型分層,除去新生兒患者,呼吸系統(tǒng)感染、中樞神經(jīng)系統(tǒng)感染及其他感染兒童患者萬古霉素血藥濃度觀測值及預(yù)測值計算獲得MPE(%)和MDPE(%)均在±20%之內(nèi),RMSE在30%之內(nèi)(呼吸系統(tǒng)感染患者為31.1%)。表明該模型的預(yù)測性能較好,可以在臨床用于模型引導(dǎo)的精準(zhǔn)用藥。觀測值與個體預(yù)測值的Bland-Altman一致性評價情況見圖2,顯示模型觀測值與預(yù)測值一致性較好,除了新生兒組偏差較大外,其他分組均顯示較好的一致性。
3 討論
基于合適的群體藥物動力學(xué)模型,結(jié)合一個或者多個血藥濃度觀測值,通過貝葉斯后驗估算個體藥動學(xué)參數(shù),以此預(yù)測穩(wěn)態(tài)下的谷濃度或AUC,目前已經(jīng)成為臨床優(yōu)化給藥方案、指導(dǎo)個體化用藥的新方法。近年來,隨著定量藥理軟件的發(fā)展和定量藥理專業(yè)知識的普及推廣,針對于某種藥物在不同患者人群中的群體藥物動力學(xué)模型被報道得越來越多[14]。但是建立的群體藥物動力學(xué)模型的結(jié)構(gòu)、納入的協(xié)變量以及患者人群的不同導(dǎo)致模型之間存在較大的異質(zhì)性[14]。在選擇合適的用于指導(dǎo)個體化用藥的群體藥物動力學(xué)模型時,通過外部驗證來評估模型的預(yù)測性能顯得至關(guān)重要。本研究通過回顧性收集兒童患者萬古霉素臨床常規(guī)血藥濃度監(jiān)測數(shù)據(jù),計算MPE(%)、MDPE(%)、RMSE等統(tǒng)計指標(biāo)[14],繪制圖形,綜合評價前期在兒童皮膚軟組織和骨關(guān)節(jié)感染的患者中建立的萬古霉素群體藥物動力學(xué)模型在其他兒童群體中的預(yù)測性能。結(jié)果表明,該模型可以外推至嬰幼兒、兒童及青少年患者用于模型引導(dǎo)的精準(zhǔn)用藥,指導(dǎo)萬古霉素在兒童患者中的個體化應(yīng)用。
截至目前,在不同的兒童和成人患者人群中已經(jīng)建立了較多的萬古霉素群體藥物動力學(xué)模型,并且也有學(xué)者對這些模型進(jìn)行了外部驗證[13,15-19]。但是納入的人群特征不同、樣本量以及每個個體的觀測值數(shù)量的差異以及建模和外部驗證數(shù)據(jù)集的取樣時間是否一致均會對外部驗證的結(jié)果造成較大的影響[20]。通常來說,建模數(shù)據(jù)集與外部驗證數(shù)據(jù)集的特征相似度越高,該群體藥物動力學(xué)模型預(yù)測結(jié)果可能越好。但是Heus團(tuán)隊[19]研究發(fā)現(xiàn),基于萬古霉素持續(xù)輸注和(或)間歇輸注數(shù)據(jù)建立的萬古霉素群體藥物動力學(xué)模型對非ICU成人患者持續(xù)輸注萬古霉素的血藥濃度結(jié)果具有低偏倚、高精確度的預(yù)測性能。該研究表明外部驗證數(shù)據(jù)與建模數(shù)據(jù)之間的相似性可能非常重要,但并不是決定該模型是否能外推用于模型引導(dǎo)的精準(zhǔn)用藥的唯一標(biāo)準(zhǔn)。盡管本研究前期建立的萬古霉素群體藥物動力學(xué)模型是在皮膚軟組織和骨關(guān)節(jié)感染的兒童患者中建立的,但外部驗證結(jié)果表明,該模型可用于我中心除新生兒外的兒童患者臨床萬古霉素個體化給藥方案制定。在新生兒群體中的預(yù)測結(jié)果呈現(xiàn)出了較大的偏差和不精確性,可能是因為新生兒患者中萬古霉素的群體藥物動力學(xué)模型通常包括矯正胎齡作為協(xié)變量,但我們前期在建立萬古霉素PopPK時,納入新生兒數(shù)量過少,未能納入該協(xié)變量的緣故[21-23]。
最大后驗貝葉斯估計方法又稱為后驗預(yù)測,是指結(jié)合一個或多個觀測濃度估計獲得個體預(yù)測值,目前已廣泛應(yīng)用于群體藥物動力學(xué)模型的外部驗證中。在本研究中,通過計算個體預(yù)測值和觀測值的MPE(%)、MDPE(%)和RMSE等統(tǒng)計指標(biāo)對模型進(jìn)行外部驗證。結(jié)果表明,萬古霉素群體藥物動力學(xué)模型在后驗情景下結(jié)合一個或兩個先驗濃度的預(yù)測性能符合臨床臨床可接受的標(biāo)準(zhǔn)[14]。也就是說,本文建立的萬古霉素PopPK模型可以結(jié)合臨床常規(guī)萬古霉素治療藥物監(jiān)測數(shù)據(jù)作為先驗值,在后續(xù)的給藥方案優(yōu)化中發(fā)揮重要作用。
本研究仍然存在一定的局限性。本研究回顧性收集萬古霉素臨床常規(guī)血藥濃度監(jiān)測數(shù)據(jù)進(jìn)行外部驗證,該數(shù)據(jù)大部分為穩(wěn)態(tài)谷濃度數(shù)值。但是需要注意的是,在實際的臨床實踐中,由于護(hù)士采血可能存在一定的誤差,萬古霉素血藥濃度樣本采集可能并未在萬古霉素給藥前30 min內(nèi),導(dǎo)致觀測值與預(yù)測值之間產(chǎn)生了偏差。此外,由于前期建立群體藥物動力學(xué)模型納入的患者腎功能均在正常范圍內(nèi),一些腎功能指標(biāo)例如血清肌酐和腎小球濾過率在前期并未納入最終模型。因此,加入腎功能描述符是否能提高人群預(yù)測性能,還需進(jìn)一步研究驗證。
4 結(jié)論
通過本研究證實,前期建立的兒童患者萬古霉素群體藥物動力學(xué)模型可以外推至嬰幼兒、兒童及青少年患者中用于模型引導(dǎo)的精準(zhǔn)用藥,從而指導(dǎo)萬古霉素在兒童患者中的個體化應(yīng)用。
參 考 文 獻(xiàn)
Liu C, Bayer A, Cosgrove S E, et al. Clinical practice guidelines by the infectious diseases society of america for the treatment of methicillin-resistant Staphylococcus aureus infections in adults and children[J]. Clin Infect Dis, 2011, 52(3): e18-55.
L?wdin E, Odenholt I, Cars O. In vitro studies of pharmacodynamic properties of vancomycin against Staphylococcus aureus and Staphylococcus epidermidis[J]. Antimicrob Agents Chemother, 1998, 42(10): 2739-2744.
Rybak M J. The pharmacokinetic and pharmacodynamic properties of vancomycin[J]. Clin Infect Dis, 2006, 42(Suppl 1): S35-39.
Rybak M J, Le J, Lodise T P, et al. Therapeutic monitoring of vancomycin for serious methicillin-resistant Staphylococcus aureus infections: A revised consensus guideline and review by the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the Society of Infectious Diseases Pharmacists[J]. Am J Health Syst Pharm, 2020, 77(11): 835-864.
Rybak M J, Le J, Lodise T P, et al. Therapeutic monitoring of vancomycin for serious methicillin-resistant Staphylococcus aureus infections: A revised consensus guideline and review by the American Society of Health-system Pharmacists, the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the Society of Infectious Diseases Pharmacists[J]. Clin Infect Dis, 2020, 71(6): 1361-1364.
He N, Su S, Ye Z, et al. Evidence-based guideline for therapeutic drug monitoring of vancomycin: 2020 update by the division of therapeutic drug monitoring, Chinese pharmacological society[J]. Clin Infect Dis, 2020, 71(Suppl 4): S363-S371.
Fuchs A, Csajka C, Thoma Y, et al. Benchmarking therapeutic drug monitoring software: A review of available computer tools[J]. Clin Pharmacokinet, 2013, 52(1): 9-22.
Cheng Y, Wang C Y, Li Z R, et al. Can population pharmacokinetics of antibiotics be extrapolated? implications of external evaluations[J]. Clin Pharmacokinet, 2021, 60(1): 53-68.
Baklouti S, Marolleau S, Chavanet P, et al. Why is it desirable to do the external evaluation of a population pharmacokinetic model?[J]. Antimicrob Agents Chemother, 2022, 66(1): e0149321.
Lv M, Yang P, Zhang S, et al. Population pharmacokinetics and dosage optimization of vancomycin in pediatric patients with skin and soft tissue infections, bone, and joint infections[J]. Antimicrob Agents Chemother, 2023, 67(1): e01624-22.
Aljutayli A, El-Haffaf I, Marsot A, et al. An update on population pharmacokinetic analyses of vancomycin, part II: In pediatric patients[J]. Clin Pharmacokinet, 2022, 61(1): 47-70.
Wang S, Yin Q, Yang M, et al. External evaluation of population pharmacokinetic models of methotrexate for model-informed precision dosing in pediatric patients with acute lymphoid leukemia[J]. Pharmaceutics, 2023, 15(2): 569-583.
Guo T, van Hest R M, Roggeveen L F, et al. External evaluation of population pharmacokinetic models of vancomycin in large cohorts of intensive care unit patients[J]. Antimicrob Agents Chemother, 2019, 63(5): e02543-18.
Lv M, Zhang S. Comment on: “External evaluation of population pharmacokinetic models for precision dosing: Current state and knowledge gaps”[J]. Clin Pharmacokinet, 2023, 62(8): 1183-1185.
Liu Y X, Wen H, Niu W J, et al. External evaluation of vancomycin population pharmacokinetic models at two clinical centers[J]. Front Pharmacol, 2021, 12: 623907.
Zhao W, Kaguelidou F, Biran V, et al. External evaluation of population pharmacokinetic models of vancomycin in neonates: The transferability of published models to different clinical settings[J]. Br J Clin Pharmacol, 2013, 75(4): 1068-1080.
Zhou Y, Long E, Shi T, et al. External validation of vancomycin population pharmacokinetic models in ten cohorts of infected Chinese patients[J]. J Glob Antimicrob Resist, 2022, 30: 163-172.
Lv C, Lu J, Jing L, et al. Systematic external evaluation of reported population pharmacokinetic models of vancomycin in Chinese children and adolescents[J]. J Clin Pharm" Therapeut, 2021, 46(3): 820-831.
Heus A, Uster D W, Grootaert V, et al. Model-informed precision dosing of vancomycin via continuous infusion: a clinical fit-for-purpose evaluation of published PK models[J]. Int J Antimicrob Agents, 2022, 59(5): 106579.
El Hassani M, Marsot A. External evaluation of population pharmacokinetic models for precision dosing: Current state and knowledge gaps[J]. Clin Pharmacokinet, 2023, 62(4): 533-540.
Jarugula P, Akcan-Arikan A, Munoz-Rivas F, et al. Optimizing vancomycin dosing and monitoring in neonates and infants using population pharmacokinetic modeling[J]. Antimicrob Agents Chemother, 2022, 66(4): e0189921.
Alsultan A, Al Munjem M F, Atiq K M, et al. Population pharmacokinetics of vancomycin in very low birth weight neonates[J]. Front Pediatr, 2023, 11: 1093171.
Hughes J H, Tong D M H, Faldasz J D, et al. Evaluation of neonatal and paediatric vancomycin pharmacokinetic models and the impact of maturation and serum creatinine covariates in a large multicentre data set[J]. Clin Pharmacokinet, 2023, 62(1): 67-76.
文章編號:1001-8689(2024)09-1051-06
收稿日期:2023-10-12
基金項目:河南省科技發(fā)展計劃(科技攻關(guān))(No. 202102310396)
作者簡介:呂萌,男,生于1990年,碩士,主管藥師,主要從事兒童個體化藥物治療研究,E-mail: lvmengyh@126.com