Lung microbiome for screening imminent exacerbation in CF

Cystic Fibrosis (CF) is a genetically borne disease that is characterized by recurrent and chronic infections of the airways. Pulmonary exacerbations (PEx) are acute aggravations of the clinical condition and lead to irreversible lung function decline and eventually to a dramatically reduced life expectancy for persons with CF. If it were possible to predict PEx and apply strategic antibiotics treatment for PEx prevention, we could substantially reduce morbidity, mortality and health care costs for CF patients. Importantly, PEx prediction is also highly relevant in COPD, asthma and pneumonia. The conventional view that airway infection in CF involves a limited suite of opportunistic bacterial pathogens has given way to recognition that CF airways typically harbor complex poly-microbial communities that colonize the lung similar to other ecologic habitats. In this project we will study the dynamics and activity of the airway microbiome and how medical treatment and the human immune response shapes these. Crucially, we will use this integrated view to predict PEx. An obstacle to studying processes that lead to PEx is the availability of appropriate, prospective data. Here we can build on a tremendous data collection containing >1900 daily sputum samples, sequenced and collected together with a suite of clinical data and current medical treatment. Furthermore, selected samples will be subjected to metagenome and metabolome analysis to illuminate microbial metabolic processes that directly influence PEx. With this unique data set and our modeling approach we can investigate and interpret the frequent changes in the airway microbiome and identify major routes that lead from a clinically-silent condition to PEx. We will apply correlation networks and graph analyses for identifying microbiome states that constitute right before PEx and deploy them for PEx prediction. Such an assessment scheme can readily be implemented in routine clinical settings. To establish the temporal context of the different processes we will implement a novel method that measures temporal sequence in correlation networks based on big data analysis. This will enable a causal association of how treatment shapes microbiome response and allow the development of strategic, model-based treatment strategies that will enhance patient prognosis. To make our research accessible to clinical use, we will implement a public online resource that can assist in treatment decisions. Of note is that we aim not alone at predictive understanding of complex processes involved in CF progression, but generate transferable concepts, which can be directly applied to other complex poly- microbial infections (e.g. COPD). With our innovative, computational approach, we will reclaim principally static high-throughput data for dynamic modeling, simulation and the prediction of individual patient trajectories.