Timely diagnostics for fungal infections are sorely needed to guide effective therapy. Invasive fungal infections are increasing in prevalence, causing millions of deaths each year worldwide, and drug resistance poses a rising threat. Due in large part to slow, outmoded diagnostics that require days of culture to identify the pathogen and report its antifungal susceptibility profile, mortality from invasive fungal infections can exceed 40%. This in turn leads clinicians to rely on empiric and prophylactic use of antifungals that may be ineffective, cause needless toxicity, and select for resistance. Rapid precision diagnostic assays are critically needed to improve patient outcomes and guide efficient deployment of our limited antifungal arsenal. To address this urgent public health need, in response to a specific funding opportunity announcement on ?Advancing Development of Rapid Fungal Diagnostics? (PA-19-080), this proposal describes a strategy for rapid fungal identification and antifungal susceptibility testing based on RNA signatures. This approach relies on a novel paradigm for pathogen diagnostics, recently validated in bacteria and implemented on a simple, robust, quantitative, multiplexed fluorescent hybridization assay on the NanoString platform. Detection of highly abundant, conserved ribosomal RNA (rRNA) sequences enables broad-range, ultrasensitive pathogen identification. Meanwhile, quantifying key messenger RNA levels following antimicrobial exposure enables phenotypic antimicrobial susceptibility testing (AST), relying on the principle that cells that are dying or growth- arrested are transcriptionally distinct within minutes from those that are not (Bhattacharyya et al, Nature Medicine, in press). Because this approach to AST measures gene expression as an early phenotypic change in susceptible strains, it does not rely on foreknowledge of the genetic basis of resistance in order to classify susceptibility, and can thus be generalized to any pathogen-antimicrobial pair. This proposal aims to first computationally design and experimentally validate a set of hybridization probes to uniquely recognize the 18S and 28S rRNA from each of 48 clinically significant fungal pathogens that together cause the vast majority of invasive fungal infections in humans. Preliminary data show that these rRNA targets are abundant enough to detect a single fungal cell without amplification, enabling ultrasensitive detection in <4 hours directly from clinical samples. Next, RNA-Seq will be used to profile transcriptional changes in 12 common fungal pathogens for which resistance has important clinical consequences in response to treatment with the three major classes of antifungals. Antifungal-responsive transcripts that best classify fungal isolates as susceptible or resistant will be chosen by adapting machine learning algorithms that were developed for this purpose in bacteria. Finally, both approaches will be piloted on simulated and real clinical fungal samples. Preliminary data suggest that these approaches can identify fungi within <4 hours from a primary sample, and deliver AST results within <6 hours of a positive fungal culture.
Fungal infections are common, increasing in frequency and severity, and cause considerable morbidity and mortality. Early recognition of invasive fungal infection, and prompt initiation of effective antifungal therapy, are key to improving patient outcomes. This proposal aims to adapt a novel paradigm for rapid pathogen identification and antimicrobial susceptibility testing through the quantitative detection of RNA signatures to transform fungal diagnostics.