Validation and early development of a blood-based rapid diagnostic test for sepsis endotypes

摘要 脓毒症是一种全身性炎症状态，作为感染的并发症发生，在严重的情况下可能会 与急性和危及生命的器官功能障碍有关。仅在美国，败血症就夺走了约27万人的生命 每年约170万确诊患者中的一年，在全球范围内，与脓毒症相关的死亡率估计为 占总死亡人数的20%。尽管脓毒症治疗方案的改进和及时 抗生素的使用降低了死亡率，没有一种药物治疗败血症是成功的。 尽管进行了100多项干预试验，但它仍被推向市场，今天被视为60年前的产品：抗生素和 支持性护理。败血症是不寻常的，因为大量的免疫反应实际上是适应性的(有益的)。 VS致病(适应不良)，因此逆转炎症反应可以导致感染消失 未选中。我们认为，一种治疗方法应该逆转适应不良(器官损害)的成分 免疫反应，同时保持适应(抗击病原体)组件的完整性。我们相信，要实现 要取得有效的进展，就需要利用免疫亚群的新方法。在这个项目中，我们建议 一种精确的医学方法将脓毒症患者细分为可治疗的亚类或“内型” 使用配套的诊断测试HostDx-Endotype。我们预计，将脓毒症患者归类为 内型可能使我们能够确定改进的治疗方案，导致发现新的靶点或 内型特异性治疗和/或现有药物再利用的途径。 在第一阶段的工作中，我们已经通过确定3种脓毒症内型(炎症性、 适应性和凝固性)，并跨多个外部数据集进行验证。我们已经证明了我们的HostDx 内型可以将患者分成新的、潜在的可治疗亚组，跨越3个出版物，使用几个 回溯性队列和两个预期队列。我们还展示了这一技术的临床应用。 基于HostDx看护点测试的InSepTM范例，这是我们对急性传染病的强大分类器 系统平台。 为了将我们的概念验证工作与HostDx-Endotype的产品开发联系起来，我们将(目标1)使用 严格的机器学习算法和流程，以最终确定并锁定最佳分类器；(目标2)应用 药物再利用的创新方法为临床确定和排序内型特异性候选药物 脓毒症治疗的研究，以及(目标3)将最终的mRNA板翻译到Inflammatx的qLAMP盒和 站台。在第二阶段完成时，我们将一起生产HostDx Endotype研究就绪型墨盒 具有用于预期临床试验的改变用途的药物的清单，并将准备进入正式产品 开发阶段；我们将启动FDA的预提交。最终，该产品的目标将是 明确地将脓毒症患者的内型与治疗干预联系起来，使更好的治疗选择成为可能 和资源分配。

ABSTRACT Sepsis, a systemic inflammatory condition, occurs as a complication of infection and in severe cases may be associated with acute and life-threatening organ dysfunction. In the US alone, sepsis claims ~270,000 lives each year among ~1.7 million patients diagnosed annually, and globally, sepsis-related mortality is estimated to account for 20% of total deaths. Although improvements in the treatment protocol for sepsis and timely administration of antibiotics have reduced mortality rates, not a single drug for sepsis has ever been successfully brought to market, despite over 100 interventional trials, and it is treated today as 60 years ago: antibiotics and supportive care. Sepsis is unusual in that a great deal of the immune response is actually adaptive (beneficial) vs. disease-causing (maladaptive), so reversing the inflammatory response can cause the infection to go unchecked. We believe that a treatment should reverse the maladaptive (organ-damaging) components of the immune response, while keeping the adaptive (pathogen-fighting) components intact. We believe that to achieve impactful progress, new approaches are required that harness immune sub-groups. In this project, we propose a precision medicine approach to subdivide sepsis patients into treatable subclasses or “endotypes” using a companion diagnostic test, HostDx-Endotypes. We anticipate that classifying sepsis patients into such endotypes may allow us to identify improved treatment regimens, leading to the discovery of new targets or pathways for endotype-specific therapies and/or repurposing of available drugs. In Phase 1 work we already demonstrated proof-of-concept by identifying 3 sepsis endotypes (Inflammopathic, Adaptive, and Coagulopathic) and validated across multiple external data sets. We have shown that our HostDx Endotypes can stratify patients into novel, potentially treatable subgroups across 3 publications using several retrospective cohorts, and two prospective cohorts. We have also demonstrated the clinical utility of this paradigm with InSepTM, our robust classifier for acute infectious disease, based on our HostDx point of care test system platform. To bridge our proof-of-concept work to product development for HostDx-Endotypes, we will (Aim 1) employ rigorous machine learning algorithms and processes to finalize and lock an optimal classifier; (Aim 2) apply an innovative approach of drug repurposing to identify and rank-order endotype-specific drug candidates for clinical studies on sepsis treatment, and (Aim 3) translate the final mRNA panel to Inflammatix’s qLAMP cartridge and platform. At Phase II completion we will have produced the HostDx Endotypes research-ready cartridge, together with a list of repurposed drugs for intended clinical trials, and will be ready to enter the formal product development phase; we will have initiated the FDA pre-submission. Ultimately, the goal of this product will be to definitively link a patient sepsis endotype with a therapeutic intervention, to enable better therapy selection and resource allocation.