Host and Microbial Biomarkers Related to the Development of Complicated Clostridium difficile Infection

 描述（由申请人提供）：美国每年有近 300 万例腹泻和结肠炎病例与艰难梭菌感染 (CDI) 有关。 CDI，不成比例地 影响老年人，可能导致暴发性、危及生命的结肠炎，并导致多次复发。迫切需要基于经过验证的生物标志物的临床决策工具来预测哪些患者将经历 CDI 的不良后果。关于哪些源自宿主、微生物组和病原体的已知和尚未发现的生物标志物能够最好地预测并发症和复发，目前还存在知识空白。我们的长期目标是开发和验证 CDI 后不良后果的准确风险预测模型，可用于指导治疗。实现这一目标的下一步以及我们拟议研究的总体目标是发现新的候选生物标志物，并确定哪些生物标志物能够在调整后的模型中最好地预测复杂的 CDI。我们的中心假设是，与仅基于临床变量的模型相比，基于生物标志物的模型能够更好地预测复杂的 CDI。为了解决这一假设，我们提出了三个具体目标：1）验证先前显示与复杂 CDI 相关的宿主生物标志物； 2）发现复杂CDI的新型微生物生物标志物； 3) 开发一个基于候选生物标志物的复杂 CDI 预测模型。我们将前瞻性地收集一组老年人的血清和粪便，以确定微生物组的特征、艰难梭菌的微生物特征、 抗毒素抗体和/或炎症介质的水平与并发症相关。将使用线性回归和机器学习技术构建多变量模型，并使用模型诊断来生成最终的便携式、基于生物标记的预测模型，以供未来研究使用。

 DESCRIPTION (provided by applicant): Clostridium difficile infection (CDI) is implicated in nearly 3 million cases of diarrhea and colitis in the U.S. each year. CDI, which disproportionately affects older adults, can result in fulminant, life-threatening colitis and lead to multiple recurrnt episodes. There is an urgent need for validated biomarker-based clinical decision-making tools to predict which patients will experience adverse outcomes from CDI. There is a gap in knowledge regarding which known and yet to-be-discovered biomarkers, derived from the host, microbiome and pathogen, will best predict complications and recurrence. Our long-term goal is to develop and validate accurate risk-prediction models for adverse outcomes following CDI that can be used to guide therapy. The next step in pursuit of that goal, and our overall objective for the proposed research, is to discover new candidate biomarkers and determine which ones together best predict complicated CDI in an adjusted model. Our central hypothesis is that a biomarker-based model will better predict complicated CDI compared to models based on clinical variables alone. To address this hypothesis we propose three specific aims: 1) validate host biomarkers previously shown to associate with complicated CDI; 2) discover novel microbial biomarkers for complicated CDI; and 3) develop a candidate biomarker-based predictive model for complicated CDI. We will collect sera and stool prospectively in a cohort of older adults to determine if characteristics of the microbiome, microbial features of C. difficile, levels of antitoxin antibodies, and/or inflammatory mediators associate with complications. Multivariable models will be constructed using linear regression and machine learning techniques and model diagnostics will be used to generate a final portable, biomarker-based predictive model for use in future studies.