POPULATION PHARMACOKINETICS/DYNAMICS--STATISTICAL ISSUES

描述(改编自申请者摘要)：本申请书共三份 处理研究的设计和分析的具体目标 群体药代动力学(PK)和药效学(PD)及其应用 癌症化疗。 第一个具体目标建议在早期的贝叶斯分析的基础上进行扩展 作为癌症和白血病B组的一部分收集的血液学数据 (CALGB)第一阶段研究。特别是，建议开发一种贝叶斯方法 半参数模型，一个足够灵活的模型，允许异质性和 患者特定参数分布的过度离散性 (随机效应分布)以及非参数回归 特定于患者的协变量。扩展模块还将包括 多个纵向结果。调查人员将重新分析CALGB 第一阶段研究和评估与申请者早期的不同之处 分析。 第二个具体目标是开发一种预测模型，与 血液毒性对患者的影响特征。调查人员将 开发一个贝叶斯分层元模型，并将其应用于来自两个 已完成的CALGB研究：目标一中提到的第一阶段研究和大型 II期乳腺癌患者辅助化疗的I期试验 癌症。 第三个目标是提出解决优化设计问题的新方法。 围绕PK/PD模型构建，诚实地考虑到 PK/PD模型中的估计和预测。尤其是， 调查人员将应用决策理论考虑制定一项 合理选择人群中患者血浆采样时间的策略 药代动力学和药效学研究。他们将评估潜力 与当前有限的采样策略相比，通过模拟实现了节省 不同假定药代动力学和药效学模型下的研究 在文献中有报道。

DESCRIPTION (Adapted from applicant's abstract): This application has three specific aims dealing with the design and analysis of studies into population pharmacokinetics (PK) and pharmacodynamics (PD), with application to cancer chemotherapy. The first specific aim proposes expanding upon an earlier Bayesian analysis of hematology data collected as part of a Cancer and Leukemia Group B (CALGB) Phase I study. In particular, it is proposed to develop a Bayesian semiparametric model, one flexible enough to allow heterogeneity and overdispersion in the distribution of patient-specific parameters (random-effects distribution), as well as non-parametric regression on patient-specific covariates. Extensions will also include models for multiple longitudinal outcomes. The investigators will reanalyze the CALGB Phase I study and evaluate differences from the applicant's earlier analysis. The second specific aim is to develop a predictive model relating hematologic toxicity to patient characteristics. The investigators will develop a Bayesian hierarchical metamodel and apply it to data from two completed CALGB studies: the Phase I study mentioned in aim one and a large Phase I trial of adjuvant chemotherapy for women with stage II breast cancer. The third aim proposes new methodology for solving optimal design problems built around PK/PD models, honestly accounting for uncertainty in the estimation and prediction in the PK/PD models. In particular, the investigators will apply decision-theoretic considerations to develop a rational strategy for picking times to sample patient plasma in population pharmacokinetic and pharmacodynamic studies. They will evaluate potential savings, compared to current limited sampling strategies, via simulation studies under various presumed pharmacokinetic and pharmacodynamic models reported in the literature.