Development of a multi-RNA signature in blood towards a rapid diagnostic test to robustly distinguish patients with acute myocardial infarction

开发血液中的多 RNA 特征以进行快速诊断测试，以强有力地区分急性心肌梗死患者

• 批准号: 10603548
• 负责人: Timothy E Sweeney
• 金额: $ 29.98万
• 依托单位: INFLAMMATIX, INC.
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-07-01 至 2024-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10603548
• 关键词: Accident and Emergency department; Acute myocardial infarction; Address; Affect; Bayesian Method; Biological Markers; Blood; Blood specimen; Chest Pain; Clinical; Clinical Sensitivity; Cohort Analysis; Collaborations; Data; Data Set; Development; Diagnosis; Diagnostic; Diagnostic tests; Electrocardiogram; Emergency Department patient; Emergency department visit; Engineering; Gene Expression; Generations; Genes; Geography; Goals; Grant; Health Care Costs; Healthcare; Heterogeneity; Immune response; Letters; Libraries; Life; Logistic Regressions; Machine Learning; Marketing; Measurement; Measures; Messenger RNA; Meta-Analysis; Methods; Modeling; Myocardial Infarction; Outcome; Patient Triage; Patient-Focused Outcomes; Patients; Performance; Phase; Phenotype; Population; Preparation; Process; Prospective Studies; RNA; Rapid diagnostics; Research; Retrospective cohort; Sampling; Sensitivity and Specificity; Sepsis; Small Business Innovation Research Grant; System; Target Populations; Testing; Time; Training; Translating; Triage; Troponin; Validation; Work; acute infection; acute symptom; analysis pipeline; bioinformatics pipeline; blind; clinical diagnostics; clinically actionable; clinically relevant; cohort; combat; data integration; genetic signature; heterogenous data; improved; instrument; machine learning classifier; machine learning method; machine learning model; machine learning pipeline; molecular diagnostics; multilayer perceptron; peripheral blood; personalized diagnostics; point of care; point of care testing; point-of-care diagnostics; product development; prototype; real world application; research clinical testing; response; risk stratification; success; support vector machine; transcriptome sequencing; transcriptomics

ABSTRACT Chest pain, the main symptom of acute myocardial infarction (AMI), accounts for ~5% of all emergency department (ED) visits. In the absence of ECG abnormalities, diagnostic gold standard for AMI relies on serial troponin (cTn) measurements which are inconclusive in 20-40% of patients, requiring additional testing and prolonged observation in the ED. A missed diagnosis of AMI without proper treatment is life threatening and thus rule-out diagnosis requires very high sensitivity. Hence, a rapid point of care (POC) test utilized as an adjunct to cTn with enhanced diagnostic performance would be revolutionary for risk stratification and timely and safe triaging of patients with suspected MI in ED. Inflammatix is a molecular diagnostics company focused on developing and bringing to market best in class, immune response based, data-driven testing. We developed a point-of-care instrument, MyRNA™, capable of quantitating up to 64 mRNAs in under 30 minutes (with <2 minutes operator time), directly from patients’ blood, in a fully disposable cartridge. We specialize in use of state-of-art multi-cohort analysis and machine learning (ML) to identify and validate robust biomarkers that generalize across real-world data heterogeneity, in diverse clinical contexts. Previous work demonstrated the potential of blood gene expression as a biomarker for MI, however a clinical test based on immune response in blood gene expression is yet to be developed. We applied our analytical framework to 6 publicly available datasets and identified a multi-gene AMI signature in peripheral blood that allows us to differentiate patients with AMI from clinically relevant controls with AUC ~ 0.95. In this project, we propose to take the AMI signature from preliminary results through research and initial development stages, up to formal clinical diagnostic development. We will generate a significant amount of independent data, leverage Inflammatix ML capabilities to further refine the mRNA signature and deliver a robust classifier ready for validation in prospective studies. In Specific Aim 1, we will generate, process, and analyze RNA-seq data for 900 blood samples from retrospective cohorts closely representing the target test population. In Specific Aim 2, we will first refine, optimize, and validate the mRNA signature; and then develop a prototype ML classifier. Specifically, we will 1) integrate expression data from all cohorts while minimizing bias; 2) apply Bayesian multi-cohort framework for final gene set selection with 300 new samples; 3) develop and evaluate discriminatory performance of AMI classifier prototype; and 4) validate the AMI classifier prototype on 600 unseen samples. These steps will produce: i) a validated set of genes for AMI; ii) an integrated dataset; and iii) a classifier prototype (AUC > 0.90), ready for clinical validation via prospective studies in Phase 2 research. This test, when developed as a cartridge on Inflammatix’s POC instrument, MyRNA™, will facilitate clinician’s triaging decisions of patients with suspected MI, improve patient outcomes, and reduce healthcare costs.

抽象的 胸痛是急性心肌梗塞 (AMI) 的主要症状，约占所有急诊的 5% 在没有心电图异常的情况下，诊断 AMI 的黄金标准依赖于连续检查。 肌钙蛋白 (cTn) 测量对于 20-40% 的患者来说是不确定的，需要额外的测试和 如果没有适当的治疗，在急诊室的长时间观察会导致 AMI 的漏诊，从而危及生命。 因此，排除诊断需要非常高的灵敏度，因此，快速护理点（POC）测试被用作一种诊断方法。 具有增强诊断性能的 cTn 辅助手段对于风险分层和及时、准确的诊断将是革命性的 对急诊室疑似心肌梗死患者进行安全分类。 Inflammatix 是一家分子诊断公司，专注于开发同类最佳产品并将其推向市场， 我们开发了一种基于免疫反应的数据驱动测试，MyRNA™，能够 在 30 分钟内（操作时间小于 2 分钟）直接从患者血液中定量多达 64 个 mRNA， 我们专注于使用最先进的多队列分析和机器学习。 （ML）识别和验证稳健的生物标志物，这些生物标志物可以在不同的现实世界数据异质性中进行推广 先前的工作证明了血液基因表达作为 MI 生物标志物的潜力， 然而，我们尚未开发出基于血液基因表达免疫反应的临床测试。 我们的分析框架对 6 个公开可用的数据集进行了分析，并确定了外周血中的多基因 AMI 特征 血液使我们能够将 AMI 患者与 AUC ~ 0.95 的临床相关对照区分开来。 在这个项目中，我们建议通过研究和初步从初步结果中获取 AMI 签名。 开发阶段，直到正式的临床诊断开发，我们将产生大量的。 独立数据，利用 Inflammatix ML 功能进一步细化 mRNA 签名并提供强大的 在特定目标 1 中，我们将生成、处理和分析分类器，以便在前瞻性研究中进行验证。 来自回顾性队列的 900 个血液样本的 RNA 测序数据，密切代表目标测试人群。 在Specific Aim 2中，我们将首先完善、优化和验证mRNA签名，然后开发原型； 具体来说，我们将 1) 整合所有群组的表达数据，同时最小化偏差；2) 应用 用于最终基因集选择的贝叶斯多队列框架，包含 300 个新样本 3) 开发和评估； AMI 分类器原型的判别性能；以及 4) 在 600 上验证 AMI 分类器原型 这些步骤将产生：i) 经验证的 AMI 基因组；ii) 综合数据集； 分类器原型（AUC > 0.90），准备通过 2 期研究中的前瞻性研究进行临床验证。 当该测试在 Inflammatix 的 POC 仪器 MyRNA™ 上开发为盒时，将有助于临床医生进行分类 疑似心肌梗死患者的决策，改善患者的治疗结果，并降低医疗成本。