Differentiation of Candida auris from other pathogenic yeasts using near-infrared spectroscopy and multivariate...

Candida (Candidozyma) auris has emerged as a major public health concern due to its multidrug resistance, high mortality rates, and outbreak potential. These challenges are intensified by the difficulty of accurately identifying this species, particularly in settings with limited laboratory resources. This difficulty arises because C. auris is closely related to other yeast species, such as those within the Candida haemulonii complex. Although we previously demonstrated that near-infrared spectroscopy (NIRS) combined with multivariate analysis can discriminate C. auris from C. haemulonii stricto sensu, its performance against other clinically important yeasts had not been evaluated.

In this study, we assessed NIRS coupled with different multivariate analytical techniques as a tool for distinguishing C. auris from C. haemulonii, C. albicans, C. tropicalis, C. parapsilosis, Nakaseomyces glabrata (formerly C. glabrata), and Pichia kudriavzevii (formerly C. krusei). Each of the seven species was cultured on fifteen Sabouraud Dextrose agar plates at 37 °C. After 72 h, three isolated colonies per plate (45 colonies per species) were subjected to Fourier-transform NIR analysis, resulting in a total of 315 spectra. The spectra were preprocessed and analyzed using principal component analysis (PCA), successive projections algorithm (SPA), genetic algorithm (GA), and linear discriminant analysis (LDA) to construct classification models.

The combination of PCA, SPA, and GA with LDA achieved 100% sensitivity, specificity, and accuracy. These findings demonstrate that NIRS coupled with multivariate analysis can reliably differentiate C. auris from other medically important yeasts. The models also showed strong discriminatory capacity among the most prevalent pathogenic yeast species, reinforcing the promise of this approach as a rapid diagnostic tool for overcoming current identification challenges.