Improving predictive capacity of models for universal influenza vaccine development

人们充分认识到需要改进和更具普遍保护作用的流感疫苗。改进工作的核心是开发更能预测人类对免疫和/或感染反应的动物模型。事实上，NIAID通用流感疫苗战略计划已经强调了这一需求。虽然动物模型可能永远无法完全预测人类的反应，但了解它们的全部优点和缺点并确定用于不同目的的最佳模型是一项重大的公共卫生需求，也是我们提出的目标背后的科学前提。这些目标是建立在我们广泛使用流感动物模型的基础上的，目的是优化免疫印迹的动物模型，改进疫苗效力测试，并确定保护和增强免疫反应的免疫相关因素。我们的总体目标是提供卓越的临床前模型，以支持通用流感疫苗的开发。我们将通过三个互补和相互关联的特定目标来实现这一目标：1)在动物模型中对重复流感抗原暴露的人类血清学反应进行优化建模；2)提高雪貂流感攻毒模型的定量性质；3)确定流感病毒引起的临床症状的血清学相关性。我们参与并与niaid最近资助的人类婴儿队列研究DIVINCI合作，支持了我们实现这些目标的能力。我们将在三种动物模型中反映这些婴儿的流感抗原暴露，并比较免疫学数据集，以确定哪一种最准确地反映了人类的反应（目的1）。这种人类和动物数据集和样本的结合提供了一种创新的前进方式，并将提供一组独特的差异启动动物，用于使用原始机器学习算法（目标3）确定感染和免疫反应的新生理参数（目标2）的免疫相关性。

The need for improved and more universally protective influenza vaccines is well recognized. Central to efforts towards improvements is the development of animal models more predictive of the human response to immunization and/or infection. Indeed, this need has been highlighted by the NIAID Strategic Plan for a Universal Influenza Vaccine. While animal models may never be able to fully predict the human response, understanding their full strengths and weaknesses and identifying the optimal models for different purposes is a significant public health need and is the scientific premise behind our proposed objectives. These objectives, which are built upon our extensive use of influenza animal models, are to optimize animal modeling of immunologic imprinting, to improve vaccine efficacy testing, and to identify immune correlates of protection and boosting immune responses. Our overall goal is to provide superior preclinical models to support universal influenza vaccine development. We will achieve this goal through three complementary and interrelated specific aims, 1) optimal modeling of human serologic responses to repeat influenza antigen exposure in animal models; 2) improving the quantitative nature of the ferret influenza challenge model; and 3) defining serologic correlates of influenza virus induced clinical symptoms. Our ability to conduct these aims is supported through our participation in, and collaboration with, a recently NIAID-funded human infant cohort, the DIVINCI study. We will mirror the influenza antigen exposures of a selection of these infants in three animal models and compare immunologic data sets to identify which most accurately reflects the human response (Aim 1). This marriage of human and animal data sets and samples offers an innovative way forward and will provide a unique set of differentially primed animals with which to determine immune correlates of novel physiologic parameters of infection and immune responses (Aim 2) using original machine learning algorithms (Aim 3).