Usefulness of MALDI-TOF Mass Spectrometry for Routine Identification of Candida Species in a Resource-Poor Setting

Identification of fungal clinical isolates is essential for therapeutic management. In resource-limited settings, identification mostly relies on biochemical tests whose sensitivity and specificity are known to be insufficient for identification of closely related or newly described species. MALDI-TOF has been shown in favored countries to be a reliable and powerful tool for microorganism identification, including yeasts. The aim of this study was to compare MALDI-TOF with routine identification procedures in a resource-poor context.

A total of 734 clinical specimens (502 vaginal swabs, 147 oral swabs, 61 bronchoalveolar lavage fluids and 24 stool samples) have been tested in the mycology unit of Fann Hospital, Dakar, Senegal. Strains isolated from culture were identified by both conventional phenotypic methods (germ tube formation and biochemical panels) and MALDI-TOF Saramis/VITEK MS, bioMérieux, France. In addition to comparing the final identification, we determined the time of obtaining the results and the cost for both approaches.

Overall, 218 (29.7 %) samples were positive for Candida. MALDI-TOF MS enabled the identification of 214 of the 218 strains isolated (98.1 %) at species level. Phenotypic approach yielded identification for 208 strains (95.4 %). Congruence between the tests was observed for 203 isolates. A discrepancy was observed for one isolate identified as Candida krusei with the phenotypic approach and Candida tropicalis with the MALDI-TOF. In addition, ten isolates identified at genus level by phenotypic methods were identified as C. glabrata (n = 8), C. tropicalis (n = 1) and C. parapsilosis (n = 1) by MALDI-TOF. The turnaround time for identification was <1 h using the MALDI-TOF compared to our routine procedures (48 h). The overall cost (reagents + expendables) per isolate was at 1.35 for the MALDI-TOF MS.

MALDI-TOF clearly outperformed the diagnosis capacities of phenotypic methods by reducing the delay of results and giving accurate identification at species level. Moreover, this approach appears to be cost-effective and should be implemented especially in resource-poor context.