Abstract
Despite insights gained through experimental models, the set of bacterial genes important for human infection is unclear for many of our most threatening pathogens. Klebsiella pneumoniae is a leading cause of health care-associated infections (HAIs) and commonly colonizes hospitalized patients, but the factors that determine whether a particular isolate causes disease or remains a colonizer are poorly understood. To identify bacterial genes associated with K. pneumoniae infection, a case-control study was performed comparing infected and asymptomatic colonized patients. Comparative bacterial genomics was combined with a conditional logit model that identified patient factors differentiating cases from controls. This method identified five gene loci associated with infection after adjustment for patient factors, including a psicose sugar utilization locus that was validated as a fitness factor during mouse lung infection. These results indicate that bacterial genome-wide association studies of patients can identify loci associated with HAIs and important in infection models. IMPORTANCE Klebsiella pneumoniae is a common cause of infections in the health care setting. This work supports a paradigm for K. pneumoniae pathogenesis where the accessory genome, composed of genes present in some but not all isolates, influences whether a strain causes infection or asymptomatic colonization, after accounting for patient-level factors. Identification of patients at high risk of infection could allow interventions to prevent or rapidly treat K. pneumoniae infections.