AUTHOR=Bous Michelle , Tielsch Malte , Papan Cihan , Kaiser Elisabeth , Weber Regine , Baumbach Jörg Ingo , Becker Sören L. , Zemlin Michael , Goedicke-Fritz Sybelle TITLE=Detection of volatile organic compounds in headspace of Klebsiella pneumoniae and Klebsiella oxytoca colonies JOURNAL=Frontiers in Pediatrics VOLUME=11 YEAR=2023 URL=https://www.frontiersin.org/journals/pediatrics/articles/10.3389/fped.2023.1151000 DOI=10.3389/fped.2023.1151000 ISSN=2296-2360 ABSTRACT=Introduction

Early diagnosis of infections and sepsis is essential as adequate therapy improves the outcome. Unfortunately, current diagnostics are invasive and time-consuming, making diagnosis difficult, especially in neonatology. Novel non-invasive analytical methods might be suitable to detect an infection at an early stage and might even allow identification of the pathogen. Our aim is to identify specific profiles of volatile organic compounds (VOCs) of bacterial species.

Methods

Using multicapillary column-coupled ion mobility spectrometry (MCC/IMS), we performed headspace measurements of bacterial cultures from skin and anal swabs of premature infants obtained during weekly screening for bacterial colonization according to KRINKO. We analyzed 25 Klebsiella pneumoniae (KP) cultures on MacConkey (MC) agar plates, 25 Klebsiella oxytoca (KO) cultures on MC agar and 25 bare MC agar plates as a control group.

Results

Using MCC/IMS, we identified a total of 159 VOC peaks. 85 peaks allowed discriminating KP and bare MC agar plates, and 51 peaks comparing KO and bare MC agar plates and 6 peaks between KP and KO (significance level of p < 0.05 after Bonferroni post hoc analysis), respectively. Peaks P51 (n-Decane) and P158 (Phenylethyl Alcohol), showed the best sensitivity/specificity/ positive predictive value/negative predictive value of 99.9% each (p < 0.001) for KP. P158 showed the best sensitivity/specificity/positive predictive value/negative predictive value of 99.9% each (p < 0.001) for KO. Comparing KP and KO, best differentiation was enabled using peaks P72, P97 and P16 with sensitivity/specificity/positive predictive value/negative predictive value of 76.0%, 84.0%, 82.6%, 77.8%, respectively (p < 0.05).

Discussion

We developed a method for the analysis of VOC profiles of bacteria. Using MCC/IMS, we demonstrated that VOCs derived from bacteria are clearly distinguishable from a bare agar plate. Characteristic peaks obtained by MCC/IMS are particularly suitable for the species-specific identification and differentiation of KP and KO. Thus, MCC/IMS might be a useful tool for in vitro diagnostics. Future studies must clarify whether similar patterns of VOCs can be detected in vivo in patients that are colonized or infected with KP or KO to enable rapid and accurate diagnosis of bacterial colonization.