Bloodstream infections are caused by ~100 live bacterial cells/mL of blood. As each mL blood contains ~3x10<sup>9</sup>blood cells, measuring antibiotic susceptibilities of blood-borne pathogens typically requires multiple growth and purification cycles prior to susceptibility testing. This ~68 hour delay to appropriate treatment negatively impacts patient outcomes and increases incidence of antibiotic resistance. To address this need, we measure scattered light signatures of bacteria recovered from blood, exposed to varying concentrations of antibiotics in a commercial flow cytometer. When exposed to antibiotics concentrations near the concentration that kills them, the scattered light signatures change from those of the no-antibiotic control, enabling us to rapidly and quantitatively measure susceptibilities. By developing new robust multidimensional statistical distance approaches - probability binning signature quadratic form (PB-sQF), we have reduced time to treatment guidance from ~68 hours to &lt;8 hours. This phenotypic antimicrobial susceptibility test has been applied to the bacterial pathogens responsible for the majority of bacteremia cases in the US, enabling rapid determinations of sensitivity and resistance from a few mL of blood. These adaptive binning and statistical distance determinations are linear with true distance between any pair of samples, enabling them to be broadly applicable to large data sets of any dimensionality. This talk will describe both our Fast Antimicrobial Susceptibility Testing (i.e. “FAST”), phenomenological descriptions of the statistical approach and implementation, as well its potential to analyze genetic or other large data sets.<br/>

Rapid determinations of antibiotic susceptibility phenotypes using label-free cytometry and new adaptive statistic

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.