Metagenomic DNA sequencing for semi-quantitative pathogen detection from urine: a prospective clinical proof-of-concept study. MUS

BackgroundSemi-quantitative bacterial culture is the standard method to diagnose urinary tract infection (UTI), but culture is time-consuming and unreliable when patients are pre-treated with antibiotics. Metagenomics could increase diagnostic accuracy and speed by sequencing the microbiota and resistome directly from urine. We performed a prospective proof-of-concept study of semi-quantitative pathogen and resistome detection in urine samples, using shotgun metagenomics.MethodsConsecutive urine samples submitted to a clinical diagnostic laboratory were included. Urine samples were screened based on DNA concentration, followed by PCR-free metagenomic sequencing of randomly selected high DNA concentration, culture positive and negative samples. A diagnostic index (DI) was calculated as the product of DNA concentration and fraction of pathogen reads. We compared results with the reference test, semi-quantitative culture, by calculating the Area Under the Curve (AUC) in a receiver operating characteristic analysis. We used ResFinder and PointFinder on the raw reads for resistance gene detection and compared results to phenotypic antimicrobial susceptibility testing (AST). ResultsWe analysed 529 urine samples of which 86 were sequenced. The AUC of the DNA concentration against culture as the reference standard was 0·85 (95%CI 0·8 to 0·89). At a cut-off value of 6·0ng/ml culture-positivity would be ruled-out with an NPV of 91% (95%CI 87-93%), and the number of samples requiring sequencing reduced by 56%. The AUC of the DI was 0·87 (95%CI 0·79-0·95). A DI cut-off value of 17·2 yielded a PPV of 93% (95%CI 85-97%) and an NPV of 69% (95%CI 55-80%). Gram-positive pathogens explained 8/9 false-negative metagenomic test results. Phenotypic and genotypic AST corresponded for 76-92% depending on the antibiotic. Extended-spectrum beta-lactamase producing E. coli showed 100% correspondence. InterpretationMetagenomics can semi-quantitatively detect pathogens as part of UTI diagnostics and is promising for resistome prediction.Funding This study funded by the COMPARE Consortium (Horizon 2020 research and innovation programme, grant number 643476).