Dermatophytosis is a superficial fungal infection with an ever-increasing number of patients. Culture-based mycology remains the most commonly used diagnosis, but it takes around four weeks to identify the causative agent. Therefore, routine clinical laboratories need rapid, high throughput, and accurate species-specific analytical methods for diagnosis and therapeutic management. Based on these requirements, we investigated the feasibility of DendrisCHIP® technology as an innovative molecular diagnostic method for the identification of a subset of 13 pathogens potentially responsible for dermatophytosis infections in clinical samples. This technology is based on DNA microarray, which potentially enables the detection and discrimination of several germs in a single sample. A major originality of DendrisCHIP® technology is the use of a decision algorithm for probability presence or absence of pathogens based on machine learning methods. In this study, the diagnosis of dermatophyte infection was carried out on more than 284 isolates by conventional microbial culture and DendrisCHIP®DP, which correspond to the DendrisCHIP® carrying oligoprobes of the targeted pathogens implicated in dermatophytosis. While convergence ranging from 75 to 86% depending on the sampling procedure was obtained with both methods, the DendrisCHIP®DP proved to identify more isolates with pathogens that escaped the culture method. These results were confirmed at 86% by a third method, which was either a specific RT-PCR or genome sequencing. In addition, diagnostic results with DendrisCHIP®DP can be obtained within a day. This faster and more accurate identification of fungal pathogens with DendrisCHIP®DP enables the clinician to quickly and successfully implement appropriate antifungal treatment to prevent the spread and elimination of dermatophyte infection. Taken together, these results demonstrate that this technology is a very promising method for routine diagnosis of dermatophytosis.
Keywords: biochips; dermatophytes; in vitro diagnostic; machine-learning methods; microbial cultures; multiplex PCR; syndromic approach.