Project Details

Urine protein biomarkers for tuberculosis diagnosis in children

Robert Husson, MD

Summary

BACKGROUND: Tuberculosis (TB) in children is very difficult to confirm because of its low bacterial burden. As a result, the diagnosis of TB is delayed or missed in many children, and TB may be incorrectly diagnosed in children with similar symptoms who do not have TB. GAP: The lack of a sensitive and specific test for TB in children that can be performed in resource-limited settings, i.e. at low cost and with little or no laboratory infrastructure, is a major gap in our ability to diagnose and treat children with TB. HYPOTHESIS: The quantity of specific host proteins in urine is different for children with TB compared to symptomatic control children who do not have TB. These differences in host proteins in urine can be used to develop predictive models to differentiate children with TB from children with similar symptoms who do not have TB. METHODS: We will perform a case control study for urine protein biomarker discovery and initial validation. We will quantify the abundance of individual proteins in the urine from children with TB and from symptomatic control children in Kenya, Peru and South Africa, followed by statistical analysis to identify proteins that are present in different amounts in the urine of children with TB compared to controls. Further computational approaches will be undertaken to identify combinations of proteins (signatures) that have greater sensitivity and specificity to screen for or diagnose children with TB from those without TB. RESULTS: We have identified several proteins, each of which is moderately predictive of TB in children compared to controls (AUROC ~0.70 to 0.75). Analysis to identify combinations of proteins that will be better predictors of TB status has identified a signature of 20 proteins with an AUROC of 0.86. Additional analysis to attempt to identify a signature of proteins that achieves better performance characteristics was undertaken using machine learning methods. A 200 protein set achieved accuracy of 89%, however reducing the number of proteins substantially reduced the model accuracy. At this time, we do not anticipate further work on this project. IMPACT: The goal of this study is to identify urine protein biomarkers for TB in children, which could lead to the development of simpler and more accurate tests to screen for and/or diagnose TB in children. Improved diagnosis will lead to more appropriate treatment and better outcomes for children with TB. Though we identified sets of proteins with moderate predictive ability, we were not able to achieve high sensitivity and specificity with a small enough set of proteins to be clinically useful.