Discovery of Urinary Tract Infection Biomarkers to Predict Longevity in the Elderly
Jeffrey Henderson, M.D. Ph.D.
Elderly patients are at elevated risk of acquiring urinary tract infections (UTIs) and also suffer a higher incidence of UTI-related complications. Recurrent UTIs are associated with greater ten-year mortality in the elderly. The prognostic value of UTI diagnosis is currently compromised by the excessive false positive rates of existing diagnostics tests in the elderly, which exceed 50%. Physicians are thus greatly reliant upon subjective patient evaluations. There is an urgent need to more reliably diagnose urinary tract infections and identify patients with elevated mortality risks. We therefore propose to use metabolomic profiling methods to discover diagnostic and prognostic urinary molecules that will form the basis of an improved clinical test. Our discovery efforts will focus on identifying high-risk bacterial subtypes and discovering their corresponding urinary molecular signatures. These molecular signatures may be derived from bacterial, human, and dietary sources. The proposed work will deliver a prototype diagnostic urine test to identify true UTIs and to stratify patients based upon risk of complications.
Virulence factor network discovery.
With our collaborator (Dr. Peter Mucha, University of North Carolina), we applied network community analysis and statistical bi-clustering to the virulence factor (VF) datasets from 337 inpatient uropathogenic E.coli isolates. From this analysis, four virulence factor-defined strain types (termed communities or biclusters) emerged. These strain types were largely defined by the presence or absence of three statistically associated VF groups. The E.coli strain types exhibited strong statistical associations with antibiotic resistance and patient sex, with one predominant group associated with both male sex and antibiotic resistance and appearing frequently among bacteremic strains. Despite thousands of prior publications addressing these virulence factors, this is the first network-based approach to identify interrelationships between them and the first to identify a relationship between VF composition and patient sex. These data suggest that E.coli are capable of using multiple strategies to cause UTI. A manuscript describing this study is nearly complete.
Genetic network analysis and biochemical profiling reveal a common association with siderophores. We observed that the bacterial networks identified above correspond well to a siderophore genotype. Siderophores are secreted, disease-associated molecules defined by their ability to scavenge iron from the human host as a nutrient source. When we used mass spectrometry to compare the number of different siderophore types actually synthesized by E.coli strain collections associated with different anatomical sites, we observed a correlation between disease severity and the number of siderophore types expressed. A pilot analysis of siderophore expression by a subset of the 337 inpatient strains suggested that inactivation of one of the siderophore genes limits a strain’s ability to cause bacteremia. This finding suggests that a biochemical expression assay informed by this genetic network will identify patients colonized by high-risk E.coli.
Individual variation in urinary E.coli growth permissiveness. The prominent role of siderophores in our initial work suggested that the siderophore-binding host protein Lipocalin-2 (Lcn2) may be particularly important during urinary tract infection (UTI) in human hosts. We indeed found that Lcn2 slows the urinary growth rate of E.coli but also observed surprisingly wide variations between individuals. We developed a bacterial growth assay that could be more readily applied to a larger number of specimens (bacterial growth kinetics are labor-intensive and difficult to standardize). This assay compared overnight growth between a uropathogen and its siderophore deficient mutant. This resulted in large (several logs of live bacteria) and reproducible differences in bacterial growth between permissive and restrictive individuals. Clinicians have long hypothesized that there exist intrinsic individual differences in UTI susceptibility differences among otherwise normal patients. This study suggests one possible explanation for this.
Growth permissiveness is associated with a stereotypical group of urinary molecules. Using mass spectrometry-based metabolomic profiling with principal components analysis, we identified molecules associated with growth permissiveness. We have begun using MS/MS (tandem mass spectrometry) analysis to iteratively fine-tune the metabolomics approach, which has resulted in a larger set of molecules with improved E.coli growth associations. We further identified a role for urine pH in this patient phenotype (easily assayed at point of care), such that the level of urinary molecules together with pH could discern a group of patients with high susceptibility to E.coli growth.
HPLC purification and chemical structural analysis as described above revealed these preliminary molecules to be a group of urinary compounds that result from an interplay between a specific liver function and the action of gut bacteria on specific dietary components. Whether this is ultimately useful only to distinguish susceptible individuals or might also help refine a therapeutic approach is yet to be determined. It is currently unclear whether this process is fixed in individuals or amenable to therapeutic manipulation. Further investigation using different mass spectrometric parameters may identify even better correlative molecules and help reveal the biochemical pathways underlying this phenotype.
Final Report Abstract:
Urinary tract infections (UTIs) are a common problem among elderly patients. Here we combined multivariate mathematical analyses with traditional microbiological and biochemical techniques to identify virulence strategies among clinical E.coli isolates and individualistic metabolic contributors to antibacterial immunity in humans. Findings from these studies provide a basis for more sophisticated measures of disease risk and suggest new therapeutic strategies for at-risk patients. Read the full Final Report.