Transcriptional Profiling to Discriminate Bacterial and Non-bacterial Respiratory Illnesses

Acute respiratory infections (ARI) occur commonly throughout life and are a leading cause of antibiotic overuse. Antibiotic use is directly linked to spread of antimicrobial resistance, which is now considered to be one of the most urgent threats to global public health. In most cases of ARI antibiotics, the microbial etiology is unknown and antibiotics are administered empirically and often inappropriately. Although sensitive molecular diagnostics allow rapid diagnosis of a variety of respiratory viruses, their impact on patient management and antibiotic prescription has been modest primarily due to concern about bacterial co-infection. Sensitive and specific diagnostic tests for bacterial lung infection are currently lacking. Gene expression profiling of whole blood represents a powerful new approach for analysis of the host response during infection. Preliminary studies using microarrays indicate that viruses and bacteria trigger specific host transcriptional patterns in blood, yielding unique "bio-signatures" that may discriminate viral from bacterial infection. Although encouraging, studies to date have not produced predictive gene sets with sufficient accuracy required for use in clinical medicine. Importantly, subgroups of patients with underlying conditions, specific clinical syndromes and those with mixed viral-bacterial infections have not been resolved by gene expression signatures. It is likely that the accuracy of diagnostic predictive gene sets can be optimized by analyzing transcriptional profiles while accounting for these host and clinical factors. This project will evaluate optimal blood predictive gene signatures using RNA sequencing in adults hospitalized with ARI to distinguish bacterial and nonbacterial illness in the presence of preexisting lung disease including asthma and chronic obstructive pulmonary disease as well as for pneumonia vs. non-pneumonic syndromes. Illnesses that have adjudicated diagnoses of viral alone, bacterial alone or mixed viral-bacterial infection will be selected for RNA sequencing and data used to develop a predictive model to discriminate bacterial and nonbacterial respiratory illness. The goal of this study is to define a limited number of host predictive expression genes that can be developed into a rapid point of care diagnostic and can be used by clinicians to discriminate bacterial and nonbacterial illness to optimally manage patients presenting to the hospital with respiratory symptoms. If successful, this approach could be extended to and validated in outpatients and other age groups in the future for maximal impact on patient care and antibiotic prescription. Given the impact of the SARS-CoV-2 pandemic on ARI, we will perform a short-term sub-study applying our methodological approaches to identify correlates of disease severity specifically in COVID19 patients.