Molecular characterization of pulmonary edema: a window to an injured lung

点击翻译按钮获取中文摘要

Acute respiratory distress syndrome (ARDS) is a heterogeneous, life-threatening condition defined by poor oxygenation and bilateral pulmonary edema that carries a mortality rate >40% in most studies. Every ARDS drug trial to date has failed, perhaps because the clinical criteria defining ARDS include a substantial subset of low-risk patients. Clinical scoring systems poorly predict which ARDS patients will develop prolonged respiratory failure and are at increased risk of death. Molecular profiling of pulmonary edema fluid could serve as a window on the alveolar microenvironment, enabling identification of high-risk patients. Pulmonary edema directly reflects lung pathobiology including loss of barrier integrity, inflammation, and epithelial damage. Unfortunately, free-flowing pulmonary edema has limited clinical utility because it can only be captured in a minority of patients. Molecular characterization of pulmonary edema is now possible in every patient with ARDS because of two critical innovations. First, we discovered that edema fluid condensing in the HME (heat and moisture exchanger) filter of the ventilator circuit, usually discarded as trash, closely reflects free-flowing edema. Second, we developed a non-targeted metabolomic fingerprinting method using mass spectroscopy that characterizes molecular components in trace quantities of fluid. Together, these innovations make it possible to test the central hypothesis that molecular features of edema fluid reflect pathobiology and enable ARDS risk stratification. To address the prognostic enrichment potential of HME edema fluid, 300 early ARDS patients will be recruited at 3 US centers, with protocolized HME edema fluid collection. Aim 1 is to study whether high HME edema fluid total protein and lung-injury specific proteins predict prolonged respiratory failure ≥48h in ARDS. Aim 2 is to test whether hypermetabolic edema fluid and inflammatory lipids predict prolonged respiratory failure ≥48h in ARDS. In Aim 3, LASSO machine learning will be used to integrate all proteomic and metabolomic edema features into a prolonged mechanical ventilation classifier. The robustness of the classifier will be assessed by measuring whether it adds prognostic value to clinical ARDS severity scores, identifying how well key molecular features are reflected in plasma, and testing for replication in an independent cohort. The goal of these studies is to examine to what extent a novel and readily available lung-specific sample, obtained early in the course of ARDS, reflects biology and can predict ARDS outcomes, thus offering prognostic enrichment for future clinical trials. Successful completion of the Specific Aims offers the potential for a much-needed classification scheme to better refine, understand, and therapeutically-target ARDS.