Klebsiella pneumoniae is one of the pathogens most frequently reported in healthcare-associated infections. Therefore, it is important to develop methodologies to monitor in real time the spread in healthcare environments of K. pneumoniae strains with high potential to cause infection outbreaks. The approach currently considered the gold standard for epidemiological typing of this bacterium is whole genome sequencing (WGS), an expensive and technically challenging procedure. The IR Biotyper (Bruker Daltonics, GmbH) is a recent instrument based on Fourier-transform infrared spectroscopy (FTIR), which enables rapid, low-cost, and easy-to-perform typing of bacterial isolates. However, further studies are still needed to evaluate the effectiveness of the IR Biotyper as a typing method.

To evaluate the ability of the IR Biotyper to type K. pneumoniae based on sequence type (ST) and capsular type (KL), as well as to develop a classifier using machine learning.

Seventy-three K. pneumoniae isolates previously characterized by WGS were selected for analysis using the IR Biotyper, applying principal component analysis (PCA) for dimensionality reduction and Euclidean distance & UPGMA as the clustering method. Of these, 54 isolates were used to build the classifier and 19 for its validation.

When ST was considered for strain analysis, ST307 isolates were grouped in the same cluster as ST11 isolates. When KL was used in the analysis, clusters showed 100% agreement with the corresponding capsular type. In addition, the developed classifier was able to classify isolates according to KL with high agreement.

This study demonstrated that KL has higher accuracy for typing K. pneumoniae strains by FTIR when compared to ST. Additionally, the IR Biotyper proved to be a cost-effective method and a promising tool for classifying isolates in a few minutes, being an important tool in outbreak analysis and detection.