# Predicting the primary infection source of Escherichia coli bacteremia using virulence-associated genes

## TL;DR

This study uses genetic data to predict if urinary tract infections caused bloodstream infections from E. coli, finding some genes are strong indicators but results vary between hospitals.

## Contribution

The study introduces a novel approach using genomic data to identify the primary infection source of E. coli bacteremia.

## Key findings

- papX and papGII genes were highly predictive of UTI as the source of bacteremia, with high positive predictive values but low sensitivities.
- The predictive power of virulence-associated genes varied significantly between general and tertiary referral hospitals.
- Genomic data shows potential for source identification, but differences between hospital populations hinder clinical implementation.

## Abstract

To investigate the role of E. coli virulence-associated genes (VAGs) in predicting urinary tract infection (UTI) as the source of bacteremia in two distinct hospital populations, one with a large general catchment area and one dominated by referrals.

E. coli bacteremias identified at Department of Clinical Microbiology (DCM), Hvidovre Hospital and DCM, Rigshospitalet in the Capital Region of Denmark from October to December 2018. Using whole genome sequencing (WGS), we identified 358 VAGs from 224 E. coli bacteremia. For predictive analysis, VAGs were paired with clinical source of UTI from local bacteremia databases.

VAGs strongly predicting of UTI as primary infection source of bacteremia were primarily found within the pap gene family. papX (PPV 96%, sensitivity 54%) and papGII (PPV 93%, sensitivity 56%) were found highly predictive, but showed low sensitivities. The strength of VAG predictions of UTI as source varied significantly between the two hospital populations. VAGs had weaker predictions in the tertiary referral center (Rigshospitalet), a disparity likely stemming from differences in patient population and department specialization.

WGS data was used to predict the primary source of E. coli bacteremia and is an attempt on a new and different type of infection source identification. Genomic data showed potential to be utilized to predict the primary source of infection; however, discrepancy between the best performing profile of VAGs between acute care hospitals and tertiary hospitals makes it difficult to implement in clinical practice.

The online version contains supplementary material available at 10.1007/s10096-024-04754-6.