Methods for Long-Term Follow-Up of HIV-Infected Patients

描述（由申请人提供）：本资助申请的目标是开发了解抗逆转录病毒治疗对抗病毒耐药性发展的后果的方法。耐药性的发展是随着时间的推移获得突变的结果，这一过程既反映了病毒复制的数量，也可能反映了病毒的适应性。我们打算共同模拟这些过程，以研究病毒基因型和病毒载量或适应性以及CD4计数等纵向措施的复杂协同进化。具体而言，我们的目标是：1）在病毒遗传进展的隐马尔可夫模型中纳入时变协变量，以检查病毒多样性、病毒载量和其他因素对抗病毒耐药途径的影响。2)在已知遗传状态存在错误的情况下，联合建模耐药遗传途径（使用隐马尔可夫模型）和病毒学失败时间。3)将病毒基因型与表型相关联并研究存在其他突变时特定突变的影响的半参数方法。4)当测量时间任意时，用于病毒载量重复测量的两组比较的U统计方法。5)评估测量误差和缺失的高维纵向标记的替代性。 目的1）研究病毒载量、治疗、依从性史或其他随时间变化的因素如何影响病毒耐药的途径。目的2）研究遗传途径对纵向指标（如病毒载量或适应度）时间进程的影响。目标3）扩展了在最后一个资助期开发的方法，使基因型与任何反应变量相关。新的方法是半参数的，他们应该是更强大的非正态性的错误和更强大的比以前的方法。目的4）源于对AIEDRP概念表中提出的问题的考虑，以研究原发性耐药突变对治疗病毒学应答的影响。目的5）研究病毒载量的短期应答和获得耐药突变的信息在多大程度上可以作为长期临床结局的替代品。

DESCRIPTION (provided by applicant): The goal of this grant application is development of methods for understanding the consequences of antiretroviral treatment for the development of antiviral resistance. The development of resistance results from the acquisition of mutations over time-a process that both reflects and contributes to amount of replicating virus and perhaps also the fitness of virus. We intend to model these process jointly to study the complex co-evolution of viral genotype and such longitudinal measures as viral load or fitness, and CD4 count. Specifically our aims are: 1) Inclusion of time-varying covariates in hidden Markov models of viral genetic progression to examine the impact of viral diversity, viral load, and other factors on pathways to antiviral resistance. 2) Joint modeling of genetic pathways to resistance (using hidden Markov models) and time to virological failure, in settings where the genetic state is know with error. 3) Semiparametric methods for relating viral genotype to phenotype and to investigate the impact of specific mutations in the presence of others. 4) U-statistic approaches for two group comparisons of repeated measures of viral load, when times of measurement are arbitrary. 5) Assessing surrogacy of high-dimensional longitudinal markers subject to measurement error and missingness. Aim 1) investigates how factors like viral load, treatment, adherence history or other time-varying factors influence the pathways to resistance taken by the virus. Aim 2) investigates the influence of genetic pathway on the time course of longitudinal measures, like viral load or fitness. Aim 3) extends the methods developed in the last grant period to relate genotype to any response variable. The new methods are semi parametric; they should be more robust to nonnormality of errors and more powerful than the previous methods. Aim 4) arose from consideration of the problems raised in an AIEDRP concept sheet to investigate the impact of primary resistance mutations on the virological response to treatment. Aim 5) investigates the degree to which information on short-term response of viral load and on acquisition of resistance mutations can serve as surrogates for longer term clinical outcomes.