Abstract
Background/Objectives: Accurate determination of colistin (COL) in vitro activity against carbapenem-resistant Acinetobacter baumannii complex (CRAB) isolates remains challenging, as the reference broth microdilution (BMD) method is labor-intensive and not routinely implemented in most clinical laboratories. Semi-automated susceptibility methods for colistin in the clinical laboratory require validation. The present study evaluated the performance characteristics of the recently introduced Vitek(®)2 card AST-N440 for COL antimicrobial susceptibility testing (AST) on CRAB isolates compared with a BMD-based reference method (ComASP Colistin). Methods: A total of 176 single-patient CRAB isolates from two distinct tertiary Greek hospitals between 2024 and 2025 were included. COL susceptibility testing was performed using Vitek(®)2 AST-N440 and compared with BMD. Minimum inhibitory concentrations (MICs) were interpreted according to EUCAST breakpoints. Method performance was evaluated by calculating categorical (CA) and essential agreement (EA), sensitivity, specificity, positive and negative predictive values (PPV/NPV), and major (ME) and very major error rates (VME) according to ISO 20776-2. Results: Compared with BMD, AST-N440 showed a sensitivity of 89.6% and a specificity of 62.3%, with a PPV and NPV of 81.7% and 76.0%, respectively. The CA (80.1%) and the EA (46.0%) were below ISO acceptance criteria. The VME rate was 10.4%, and the ME rate 37.7%. Identical MIC values were observed in 25.0% of the isolates, while Vitek(®)2 reported lower and higher MIC values than BMD in 46.6% and 28.4% of isolates, respectively. Conclusions: The Vitek(®)2 AST-N440 card performed suboptimally for COL susceptibility testing in CRAB isolates. Further validation of automated systems for COL AST is needed.