Defining Immunity to Placental Malaria using a Multi-assay Predictive Model

描述（由申请人提供）：由恶性疟原虫引起的疟疾，在孕妇中尤其严重，因为感染的红细胞（IE）表达VAR2CSA，一种与滋养细胞上的硫酸软骨素a （CSA）结合的配体，导致IE在胎盘中积累。因此，炎症和病理发生，增加了自然流产、早产和低出生体重婴儿的风险。幸运的是，抗VAR2CSA抗体（Ab）显著改善妊娠结局。基于var2csa的疫苗正在设计中，目标是诱导高水平的保护性抗体，今年将开始对一种候选疫苗进行人体安全性测试。然而，没有办法确定妇女是否对胎盘疟疾免疫。具有成本效益的诊断方法可以确定妇女的免疫水平，这将使1)医生能够提供更好的产前护理，2)疫苗开发人员能够评估妇女在接种疫苗前后的免疫水平，3)政府官员能够为孕妇制定明智的卫生政策，特别是在实施干预战略导致疟疾状况不断变化的情况下。因此，我们的目标是结合血清学和功能分析来表征介导胎盘IE清除的Ab，然后利用这些数据建立统计模型，预测女性是否有足够的免疫力来1)预防胎盘病理，2)预防PM。对不同PM免疫水平的喀麦隆孕妇的档案样本将在测定Ab对VAR2CSA的不同特征（包括特异性、亲和性和功能）的检测中进行筛选。这部分研究是直接的，因为我们的实验室已经优化了检测方法。此外，我们寻求开发两种新的功能测定方法来测量Ab的能力(a)阻断VAR2CSA与CSA的相互作用，(b)防止暴露于IE的胎盘滋养细胞的激活和失调，这是一种有助于胎盘病理的机制。血清学和功能分析的结果将有助于确定保护性Ab的特征。然后，这些数据将用于建立多分析统计预测模型。将使用两种统计方法。首先，将Ab特征与其他与免疫相关的关键变量（如年龄、重力、红细胞压积、妊娠时间）结合，建立多变量logistic模型。这种方法将产生两个“用户友好”的简单风险指数（公式），基于最少数量的检测和变量，为预防病理和感染提供最佳预测。其次，采用递归划分方法，利用数据开发分类和回归树（CART）和随机森林（RF）。由此产生的二叉分类树将很容易解释，并允许更复杂的免疫途径被纳入。一旦开发出这些模型，将使用在整个妊娠期间每月收集的档案样本数据来确定在妊娠期间，这些模型在疟疾高传播和低传播环境中何时具有最佳预测价值。

DESCRIPTION (provided by applicant): Malaria, caused by Plasmodium falciparum, is especially severe in pregnant women because infected erythrocytes (IE) express VAR2CSA, a ligand that binds to chondroitin sulfate A (CSA) on trophoblasts, causing IE to accumulate in the placenta. As a result, inflammation and pathology occur, increasing the risk of spontaneous abortions, premature deliveries, and low birth weight babies. Fortunately, antibodies (Ab) against VAR2CSA significantly improve pregnancy outcomes. VAR2CSA-based vaccines are being designed with the goal of inducing high levels of protective Ab, and safety testing of one candidate vaccine will begin in humans this year. However, there is no method for determining if a woman is immune to placental malaria (PM). The availability of a cost-effective diagnostic approach that identifies a woman's level of immunity will allow 1) doctors to provide better prenatal care, 2) vaccine developers to assess the level of immunity women have before and after vaccination, and 3) government officials to make intelligent health policies for pregnant women, especially with the changing malaria landscape due to implementation of intervention strategies. Therefore, our goal is to use a combination of serological and functional assays to characterize Ab that mediate clearance of IE from the placenta and then use the data to develop statistical models that predict whether a woman has sufficient immunity to 1) prevent placental pathology and 2) prevent PM. Archival samples from pregnant Cameroonian women with different levels of immunity to PM will be screened in assays that measure different characteristics of Ab to VAR2CSA, including specificity, avidity, and function. This part of the study is straight-forward as the assays are already optimized in our laboratory. In addition, we seek to develop two new functional assays that measure the ability of Ab to (a) block the interaction of VAR2CSA with CSA and (b) prevent activation and dysregulation of placental trophoblasts exposed to IE, a mechanism that contributes to placental pathology. Results from the serological and functional assays will help identify characteristics of protective Ab. Then, th data will be used to build multi-assay statistical prediction models. Two statistical approaches will be used. First, multivariable logistic models will be built based on a combination of Ab characteristics and other key variables related to immunity (e.g., age, gravidity, hematocrit, length of gestation). This approach will result in two "user friendly" simple risk indexes (formulas) based on the least number of assays and variables that provide the optimal prediction for prevention of pathology and infection. Second, a recursive partitioning approach will be taken to develop classification and regression trees (CART) and random forests (RF) using the data. The resulting binary classification trees will be easy to interpret and allow more complex immunological pathways to be incorporated. Once the models are developed, data from archival samples collected monthly throughout pregnancy will be used to determine when during pregnancy the models have the best predictive value in high and low malaria transmission settings.