Real-time identification of carbapenemase-producing Klebsiella pneumoniae lineages and outbreak detection using FT-IR ATR

Expansion of carbapenemase-producing Klebsiella pneumoniae (CP-Kp) is driven by nosocomial dissemination, and effective infection control depends on timely and reliable typing data. Here, we evaluated our previously developed Fourier-transform infrared spectroscopy (FT-IR) with attenuated total reflectance (ATR) workflow for real-time typing of Kp capsular (KL)-types and lineages to support infection control. FT-IR spectra were acquired from Columbia agar with 5% sheep blood cultures of all CP-Kp infection isolates (n=136) from hospitalized patients at a northern Portugal hospital (April 2022 - March 2023), and analyzed using automated machine-learning (ML) classification models. Typing results were confirmed by wzi sequencing, MLST and/or WGS. FT-IR typing on Columbia agar plates showed 73% sensitivity, 79% specificity and 74% accuracy. Our method correctly typed 94% of typeable isolates (78/83), from which 87% were comunicated in <24h. Sixty percent of non-typeable isolates were considered false negatives, but the majority (66%) was correctly predicted when re-tested in Mueller-Hinton agar, improving sensitivity (92%), specificity (76%) and accuracy (89%) of Kp typing. Three Kp lineages (ST147-KL64, ST15-KL19, ST268-KL20) represented 74% of the sample, with ST268-KL20 causing an outbreak in Neonatal Intensive Care unit, quickly recognized by FT-IR enabling immediate infection control measures. Epidemiological links between patiens were mostly found on medical, surgical and urology units, using EpiLinx software. Most isolates (98%) produced KPC-3. Our FT-IR ATR ML-based typing workflow demonstrated high performance standards in real-time and high adaptability to clonal dynamics. The unprecedent time-to-response (same day of species identification) represents an opportunity to implement timely and effective infection control measures. ImportanceThis study represents the first prospective and real-time evaluation of FT-IR spectroscopy to type multidrug resistant Klebsiella pneumoniae to support surveillance and infection control. We demonstrate a high sensitivity, specificity and accuracy of a previously developed workflow that allows precise identification of K. pneumoniae lineages. The adaptability to changes in clonal dynamics and bacterial typing in <24h offer significant advantages in both high- and low-income countries for a timely infection control and improvement of antimicrobial resistance management.