Drug Effect Discovery Through Data Mining and Integrative Chemical Biology

描述(申请人提供)：小分子药物是现代医学实践的基石。然而，它们的使用受到意想不到的副作用的困扰，这些副作用通常只在后期临床试验中或上市后才会出现。因此，出现了一些备受瞩目的药物停药事件，而新药开发却很匮乏。确定药物治疗的组合效应是特别值得关注的。在药物进入市场之前，很难对这些相互作用进行实证研究，因为在大多数晚期临床药物(III期)研究中，联合处方药的样本很少。一些相互作用可以根据共同的新陈代谢途径的知识来预测，但许多是特殊的，很难预测。因此，我们必须创建监测方法来检测意想不到的药物效果和相互作用，利用电子健康记录等大规模临床数据库的力量。挖掘电子健康记录数据以确定药物不良反应是一项日益重要的研究挑战。例如，根据国会的一项授权，食品和药物管理局(FDA)在2009年设立了微型哨兵倡议--这是一项试点研究，将来自31家机构的申报和管理数据联系起来，目的是监测药品安全监督。此外, 公私伙伴关系(如观察性医疗成果伙伴关系)如雨后春笋般涌现，以建立安全监测的数据管理和分析标准。然而，电子健康记录在药物监测方面的潜力也伴随着同样数量的挑战。其中许多挑战在于用于二次分析的数据的质量(或者更确切地说，是缺乏质量)。存储在EHR中的数据通常是脏的、有噪音的和丢失的。除了与数据捕获有关的问题外，这些数据还存在偏差，这混淆了分析，使数据挖掘结果难以解释。在联合治疗的背景下，这些问题变得特别尖锐，在这种情况下，暴露的患者队列通常很小，并受到未知(即未研究)偏见的影响。在这项建议中，我们提出了一种药物安全监测策略，它将最先进的信号检测算法与化学系统生物学数据相结合，目的是识别联合疗法的意外效果。我们提出了一种将定量信号检测和化学系统生物学相结合的综合方法来从大型临床数据库中挖掘药物效应。这将需要在观测统计数据挖掘、网络分析和综合化学系统生物学方面的创新。其结果将是一套发现药物效应并将其与 分子相互作用网络。这些资源将帮助联邦监管机构在人群层面更好地监测药物的安全性，希望在生理层面了解药物影响的药理学家，以及探索人类疾病新疗法的药物开发研究人员。

DESCRIPTION (provided by applicant): Small molecule drugs are the cornerstone of modern medical practice. However, their use is plagued by the onset of unexpected side effects, often seen only in late-stage clinical trials or after release to the market. As a result, there have bee a number of high profile drug withdrawals and a dearth of new drug development. Characterizing the combinatorial effects of drug treatment is of particular concern. It is very difficult to empirically study these interactions before drugs enter the market because of the small samples of co- prescribed drugs in most late stage clinical drug (Phase III) studies. Some interactions can be predicted based on knowledge of shared pathways of metabolism, but many are idiosyncratic and difficult to predict. Thus, we must create surveillance methods to detect unexpected drug effects and interactions that leverage the power of large-scale clinical databases such as the electronic health records. Mining of electronic health record data for the purpose of identifying adverse drug effects is an increasingly important research challenge. For example, in response to a congressional mandate the Food and Drug Administration (FDA) established the mini-sentinel initiative in 2009 -- a pilot study that links claims and administratve data from over 31 institutions for the purpose of monitoring drug safety surveillance. In addition, public-private partnerships (e.g. the Observational Medical Outcomes Partnership) have sprouted to establish data management and analysis standards for safety surveillance. However, the potential of the EHR for drug surveillance is paralleled by an equal number of challenges. Many of these challenges are in the quality (or rather lack thereof) of data when used for secondary analyses. Data stored in the EHR are often dirty, noisy, and missing. In addition to issues regarding data capture, these data also suffer from bias which confounds analysis and makes data mining results difficult to interpret. These issues become especially acute in the context of combination therapies where the exposed patient cohorts are often small and suffer from unknown (i.e. unstudied) biases. In this proposal we present a drug safety surveillance strategy which integrates state-of-the-art signal detection algorithms with chemical systems biology data for the purpose of identifying unexpected effects of combination therapies. We present an integrative methodology which combines quantitative signal detection and chemical systems biology to mine drug effects from a large clinical database. This will require innovations in observational statistical data mining, network analysis, and integrative chemical systems biology. The result will be a set of tools for discovering drug effects and linking them to molecular interaction networks. These resources will aid federal regulators to better monitor the safety of drugs at the population level, pharmacologists who wish to understand the effects of drugs at the physiological level, and drug development researchers to explore new treatments of human disease.