# Effects of Combined Bacterial Infection and Radiation Injury on Biofluid Metabolite Profiles in the Murine Model

## TL;DR

This study shows how bacterial infection and radiation exposure together affect metabolite profiles in mice, helping improve radiation dose assessment in real-world scenarios.

## Contribution

The study introduces a combined biofluid metabolomic model that accounts for infection and radiation exposure with high accuracy.

## Key findings

- Bacterial infection altered metabolomic signatures in a biofluid-specific manner.
- A combined serum and urine random forest model predicted radiation dose and infection status with 90% accuracy.
- Urinary metabolites showed additive effects in infected animals exposed to 6 Gy radiation.

## Abstract

Rapid biodosimetry
tools are needed to assess radiation
exposure
in scenarios complicated by secondary infections. This study evaluated
how Listeria monocytogenes infection
impacts metabolite-based biodosimetry in male C57BL/6 mice. The mice
were infected and exposed to 0, 2, or 6 Gy X-rays at 4 days postinfection.
Untargeted metabolomics was performed on serum and urine at 1 day
postirradiation. We found that the effect of bacterial infection increased
white blood cell counts and altered metabolomic signatures in a biofluid-
and compound-specific manner. Infection alone altered select serum
lipids and urinary TCA intermediates. Some urinary metabolites displayed
additive effects in infected animals exposed to 6 Gy. The best model
for combined biofluids (serum: lysophosphatidylcholines [14:0] and
[22:5], glycerophosphatidylcholines [42:8] and [42:11] and citrate;
urine: glutamic acid, creatine, propionylcarnitine, acetylspermidine,
and hexanoylglycine) was determined with a multivariate random forest
analysis model. A combined biofluid random forest model predicted
the radiation dose and infection status with 90% accuracy (RMSE =
1.31 Gy). These findings support the development of robust, multiplexed
biodosimetry panels capable of accounting for real-world confounders
like infection. Such models can improve the precision of triage decisions
following radiological emergencies (raw data available at Metabolomics
Workbench Study IDs ST004101 and ST004100).