Comparison of Five Assays for Rapid Antimicrobial Susceptibility Testing (RAST), Including Disk Diffusion, MIC-Based Systems, and Genotypic Methods: A Prospective Evaluation

Early and targeted therapy is pivotal for the outcome of bloodstream infection, a life-threatening condition with high mortality. EUCAST developed a protocol for rapid antimicrobial susceptibility testing (RAST) that allows testing directly from positive blood cultures (BCs). Recently, fully automated minimum inhibitory concentration (MIC)-based systems were introduced. This is the first study to compare two of these systems with agar-based RAST and genotypic resistance testing. Ninety-seven Gram-negative BCs were prospectively analyzed using five different RAST assays. Agar-based RAST according to EUCAST was performed using conventional disk diffusion (DD) and a total laboratory automation solution (BD Kiestra), with reads at 4, 6, and 8 h. MIC-based, fully automated RAST was performed using VITEK Reveal (bioMérieux) and dRAST (Quantamatrix). Identification and genotypic resistance testing were additionally analyzed with the Biofire BCID2 multiplex PCR (bioMérieux). A total of 6715 genotypic and phenotypic results were compared with reference methods. Biofire achieved 100% sensitivity and specificity for species identification and resistance gene detection; 0–28% of BCs were unanalyzable due to missing RAST breakpoints. Categorical agreement ranged from 82.2% (4 h DD) to 95.4% (Reveal). Agar-based RAST yielded significantly more results in the area of technical uncertainty than other methods. Kiestra-based DD outperformed conventional DD after short incubation. Reveal had the highest rate of false-susceptible results (2.4%), particularly for penicillin. dRAST tended to false-resistant results (3.4%), particularly with motile organisms like Proteus mirabilis . Automation of RAST provides more usable results than conventional DD, but each assay has specific weaknesses that users need to be aware of.