Center for Anti-Infective Research and Development, Hartford Hospital, Harford, Connecticut, USA.
Michigan State University, East Lansing, Michigan, USA.
J Clin Pharmacol. 2020 Feb;60(2):172-180. doi: 10.1002/jcph.1510. Epub 2019 Aug 18.
Monte Carlo simulations (MCSs) are used in antibiotic development to predict the probability of pharmacodynamic target attainment (PTA) for a dosing regimen. However, for β-lactam/β-lactamase inhibitor combinations (BL-BLICs), methods for linking simulated concentration profiles of the β-lactam (BL) and β-lactamase inhibitor (BLI) components are rarely described. Using a previously defined pharmacokinetic model of ceftazidime/avibactam from critically ill patients, we performed four 5000-patient MCSs using different methods of increasing complexity to couple the BL and BLI components and compared PTA for ceftazidime and avibactam targets of >70% fT>MIC and >70% fT>1 mg/L, respectively, at MICs from 1 to 128 mg/L. Method A ignored all covariates and correlations, whereas methods B, C, and D enhanced associations by adding (B) pharmacokinetic parameter correlation within each drug only; (C) pharmacokinetic parameter correlation within each drug and creatinine clearance (CRCL); and (D) pharmacokinetic parameter correlation within each drug, CRCL, and pharmacokinetic parameter correlation between drugs. Method D produced a simulated patient population that best recapitulated the observed relationships between pharmacokinetic parameters in actual patients. Ceftazidime/avibactam PTA at MIC 8 mg/L (the susceptibility break point) and 16 mg/L ranged from 92.4% to 98.3% and 80.2% to 88.4%, respectively. PTA was lowest with method A, whereas PTA estimates were similar for all other methods. Compared with ignoring all pharmacokinetic parameter associations, the inclusion of covariate relationships and parameter correlation between both components of ceftazidime/avibactam leads to fewer patients with discordant pharmacokinetic parameters and results in higher PTA. Consideration of these methodologies should guide future MCS analyses for BL-BLIC.
蒙特卡罗模拟（MCS）用于抗生素开发，以预测给药方案的药效目标达标率（PTA）。然而，对于β-内酰胺/β-内酰胺酶抑制剂组合（BL-BLIC），很少有描述将β-内酰胺（BL）和β-内酰胺酶抑制剂（BLI）成分的模拟浓度曲线联系起来的方法。我们使用先前从危重症患者中定义的头孢他啶/阿维巴坦药代动力学模型，通过不同的方法进行了四次 5000 名患者的 MCS，这些方法的复杂性逐渐增加，以将 BL 和 BLI 成分耦合起来，并比较了 MIC 为 1 至 128mg/L 时，分别>70% fT>MIC 和>70% fT>1mg/L 的头孢他啶和阿维巴坦目标的 PTA。方法 A 忽略了所有协变量和相关性，而方法 B、C 和 D 通过仅在每种药物中添加（B）药代动力学参数相关性、（C）药代动力学参数相关性在每种药物和肌酐清除率（CRCL）中以及（D）在每种药物、CRCL 和药物之间的药代动力学参数相关性来增强关联。方法 D 产生了一个模拟患者群体，该群体最好地再现了实际患者中药物动力学参数之间的观察到的关系。MIC 为 8mg/L（药敏折点）和 16mg/L 时，头孢他啶/阿维巴坦的 PTA 范围为 92.4%至 98.3%和 80.2%至 88.4%。方法 A 的 PTA 最低，而其他所有方法的 PTA 估计值相似。与忽略所有药代动力学参数关联相比，包含协变量关系和头孢他啶/阿维巴坦两者成分之间的参数相关性可导致具有不和谐药代动力学参数的患者更少，并导致更高的 PTA。考虑到这些方法学，应指导 BL-BLIC 的未来 MCS 分析。
