Statistical Methods for Long-Term HIV Dynamic Modeling and Design

描述（申请人提供）：病毒动力学研究是近年来HIV/AIDS研究中最重要的发展之一，有助于了解HIV发病机制和抗逆转录病毒（ARV）治疗方法。然而，大多数研究都集中在短期病毒动力学上，其中的模型可能不适用于长期动力学。基于生理学的数学模型和统计方法在艾滋病研究中起着至关重要的作用。在长期治疗期间，建立抗逆转录病毒（ARV）治疗的病毒学反应（VR）关系对于开发有效的治疗方法至关重要。这是一项具有挑战性的任务，因为一个实用的模型必须结合多种治疗因素，包括但不限于药物浓度、药物依从性、药物敏感性。病毒动力学可以通过微分方程来建模，但是在估计和评估这种微分方程模型的统计方法方面只有有限的发展。该项目的目标是通过具有时变系数但没有封闭形式解的微分方程系统建立病毒动力学模型，并将其应用于表征长期病毒动力学。该提案的目标1是(a)研究ARV治疗下长期HIV/ t细胞动力学和病毒学/免疫反应的模型，包括时变的药物疗效、药代动力学、药物依从性和耐药性；(b)开发灵活的方法来拟合贝叶斯非线性混合效应模型，其中包括患者之间的动态变化；这些方法的基本原理已经很好地建立起来了，但是在非线性时变系数微分方程系统的背景下，这些方法的应用仍然是创新的，但是没有封闭形式的解。目的二是将所建立的模型和方法应用于艾滋病临床试验，并通过模拟对模型进行评价。它将侧重于(a)验证模型和方法；(b)探索VR与药物浓度之间的药效学关系，这些药物浓度以药代动力学参数为特征，并结合其他混杂因素，如药物依从性和耐药性，并确定VR的关键决定因素的临床因素；(c)评估艾滋病临床试验中使用的方案设计在产生预期反应方面的有效性，从而在剂量水平、受试者人数、时间和结果监测频率方面指导抗逆转录病毒疗法的选择。这项研究将为临床因素在介导抗逆转录病毒治疗的长期有效性方面的作用提供新的视角，从而更好地了解艾滋病毒的发病机制和长期病毒学反应，并有可能在理解针对现有治疗的临床试验设计的定量评估方面取得重大进展。虽然本课题将集中研究HIV动力学，但纵向动力学系统的基本概念和本课题提出的方法一般适用于其他领域的动力学系统，如PK/PD研究、生物医学和公共卫生，只要它们符合相关的技术规范——一组微分方程。

DESCRIPTION (provided by applicant): The study of viral dynamics is one of the most important developments in recent HIV/AIDS research for understanding HIV pathogenesis and antiretroviral (ARV) therapies. However, most studies focused on short-term viral dynamics, and the models therein may not be applicable to long-term dynamics. Physiologically-based mathematical models and statistical methods play a critical role in AIDS research. Establishing the relationship of virologic responses (VR) to antiretroviral (ARV) therapy during long-term treatment is critical to the development of effective treatments. This is a challenging task because a practical model must incorporate multiple treatment factors including, but not limited to, drug concentration, drug adherence, drug susceptibility. Viral dynamics may be modeled through differential equations, but there has been only limited development in statistical methodologies for estimating and evaluating such differential equation models. The goals of this project are to develop viral dynamic models via systems of differential equations with time-varying coefficients but without a closed-form solution, and to apply them for characterizing long-term viral dynamics. Aim 1 of this proposal is to (a) study models for long-term HIV/T-cell dynamics and virologic/immunologic responses under ARV therapies, incorporating time-varying drug efficacy, pharmacokinetics, drug adherence and drug resistance; (b) develop flexible methods for fitting Bayesian nonlinear mixed-effects models which incorporate between-patient variations in dynamics; the basic principle of these methodologies were well established, but the applications of methods are nonetheless innovative within the context of a system of nonlinear differential equations of time-varying coefficient, but without a closed-form solution. Aim 2 is to apply the models and methods developed to AIDS clinical trials and to evaluate the models through simulations. It will focus on (a) validating the models and methods; (b) exploring pharmacodynamic relationships between VR and drug concentrations characterized by pharmacokinetic parameters in conjunction with other confounding factors such as drug adherence and resistance, and identifying clinical factors that are critical determinants of VR; (c) evaluating protocol designs used in AIDS clinical trials for their effectiveness in generating desired responses, therefore guiding the selection of ARV therapies with respect to level of dosing, number of subjects, timing and frequency of outcome monitoring. The proposed research will cast new lights on HIV dynamics in terms of the roles of clinical factors in mediating the long-term effectiveness of ARV therapies, hence a better understanding of HIV pathogenesis and long-term virologic responses, and will potentially lead to significant progress in understanding quantitative evaluation of clinical trial designs in response to existing therapies. Although this proposal will concentrate on HIV dynamics, the basic concept of longitudinal dynamic systems and the proposed methodologies in this project are generally applicable to dynamic systems in other fields such as PK/PD studies, biomedicine and public health as long as they meet the relevant technical specification-a set of differential equations. PUBLIC HEALTH RELEVANCE: The AIDS epidemic remains a grave public health threat world-wide. HIV dynamic studies have contributed significantly to the understanding of HIV pathogenesis and antiviral treatment strategies for AIDS patients. The overall goal of this project is to develop long-term HIV dynamic models and associated statistical methods for identifying clinical factors that are critical determinants of virological responses and for providing quantitative guidance to select and design antiretroviral treatments with respect to level of dosing, number of subjects, timing and frequency of outcome monitoring.