Combined pathogen and host-based diagnostic to identify etiology of lower respiratory tract infection

Lower respiratory tract infection (LRTI) is the leading infectious cause of death globally. Despite its prevalence, the exact etiology of LRTI is unknown in the vast majority of cases. Even when identified, bacteria or viruses in nasopharyngeal (NP) or sputum samples may be colonizers in the upper tract rather than the cause of infection in the lower tract. Unclear LRTI etiology results in the overprescription of antibacterials, which in turn drives the global crisis in antibacterial resistance. Antibacterial overprescription and resistance are greater in low- or middle-income countries (LMICs), where basic diagnostic capacity is limited. Host-based diagnostics, which assess the host immune response to infection, have recently emerged as a complementary method to pathogen-based diagnostics for identifying the class of respiratory infection. Our team has developed host- based gene expression classifiers using peripheral blood samples to differentiate viral versus bacterial respiratory infection. The goal of the current application is to develop an integrated diagnostic that uses a single, non-invasive NP sample to detect both pathogen and host response to identify LRTI etiology. The following specific aims will be conducted at a collaborative research site in Sri Lanka: 1) Develop a novel NP- based gene expression classifier to identify viral versus non-viral LRTI, and 2) Design and validate an integrated pathogen and host gene expression test to identify viral versus non-viral LRTI using a quantitative real-time polymerase chain reaction (qRT-PCR) assay. For aim 1, we will use previously collected NP samples from clinically adjudicated viral and non-viral LRTI patients in Sri Lanka and conduct low-input RNA sequencing. Machine-learning approaches will identify host gene expression classifiers that discriminate viral versus non-viral LRTI. For aim 2, the genes identified in the NP-based classifier, as well as nucleic acid targets for two respiratory viruses that are frequently implicated in true infection as well as asymptomatic colonization (SARS-COV-2 and human rhinovirus [HRV]), will be migrated onto TaqMan Low-Density Array (TLDA) cards. A prospective cohort of patients will be enrolled in Sri Lanka, and etiological testing and clinical adjudications will be performed as the reference standard to identify viral (including SARS-CoV-2 and HRV) and non-viral LRTI. Using an optimally retrained and parsimonious viral versus non-viral classifier, we will perform a feasibility analysis of incorporating pathogen detection and host-response classifier. Among samples with TLDA-based pathogen detection for SARS-CoV-2 or HRV, performance of the host- response classifier to distinguish viral versus non-viral LRTI will be assessed. Successful completion of these aims will result in the development of a novel diagnostic that integrates host and pathogen detection using a single, non-invasive NP sample to identify the etiology of LRTI. Translation of this assay to a rapid platform will help shift the current diagnostic paradigm for LRTI.