Improved Genomic Prediction of Staphylococcus epidermidis Isolation Sources with a Novel Polygenic Score

Area of Science:

• Microbiology
• Genomics
• Infectious Diseases

Background:

• Staphylococcus epidermidis causes challenging infections, including sepsis in premature infants and intravascular infections.
• Distinguishing true infections from contamination is difficult, impacting diagnosis and treatment.
• Previous methods struggle to differentiate true bacteremia from blood culture contamination.

Purpose of the Study:

• To develop and validate a polygenic score for predicting Staphylococcus epidermidis isolation sources.
• To assess the accuracy of polygenic scores in distinguishing hospital vs. community origins and infection vs. contamination.
• To evaluate the utility of genome-wide association studies in predicting complex bacterial phenotypes.

Main Methods:

• A phylogeny-informed genome-wide association study (GWAS) was conducted on 88 S. epidermidis isolates.
• Polygenic scores were calculated based on the summed effect sizes of genomic variants (accessory genes and SNPs).
• Predictive models were trained and tested using in-sample and out-sample data from diverse patient populations.

Main Results:

• Polygenic scores from accessory genes achieved high accuracy (up to 98% in-sample, 65-91% out-sample) in distinguishing hospital from community sources.
• Scores from accessory genes and SNPs reached 76% in-sample accuracy for differentiating infection from contamination.
• Model accuracy improved with more similar population structures in training and testing datasets.

Conclusions:

• This study introduces the first polygenic score approach for predicting complex bacterial phenotypes in S. epidermidis.
• The polygenic score method demonstrates significant potential for enhancing the diagnosis of S. epidermidis infections.
• Accurate source prediction is crucial for effective clinical management of S. epidermidis.