A rRNA hybridization-based approach for rapid and accurate identification of diverse fungal pathogens

Invasive fungal infections are a global threat for which early diagnosis is critical for patient outcomes, but current diagnostic measures remain notoriously slow. Here we extend our multiplexed, hybridization-based rRNA-targeted strategy for rapid, sensitive pathogen identification, previously designed for diverse bacteria and Candida species, to identify diverse fungal pathogens. We created a set of 91 probes targeting 86 medically relevant fungal species, designed to recognize regions of differential conservation across taxonomic groupings, from class- to species-specific probes. We assessed assay performance across a Training Set of 93 clinical isolates spanning 32 species of common fungal pathogens across 18 genera, with Pearson correlations of probeset reactivity profiles identifying the pathogen at the species, genus, and family level with 83%, 94%, and 95% accuracy, respectively, in a leave-one-out analysis. We developed a new classifier on this Training Set, using taxonomic categories to select progressively more informative probes at each taxonomic level. After optimization, we assessed performance on an independent Validation Set of 54 clinical isolates spanning the same species as the Training Set, with 91%, 94%, and 98% at the species, genus, and family levels, respectively. We piloted our assay on formalin-fixed paraffin-embedded (FFPE) tissue, demonstrating rapid, culture-independent fungal identification from this high-value clinical sample type, often the sole specimen available. The assay requires <30 minutes hands-on time (or <65 minutes from FFPE tissue), returning results in <8 hours from cultured specimen on an RNA detection platform available in clinical laboratories.