Multisite evaluation of blAST, a direct-from-urine phenotypic antimicrobial susceptibility test, compared with reference broth microdilution

Traditional antimicrobial susceptibility testing (AST) is time-consuming and limited by the need to isolate bacteria. The Bio-Labeling Antibiotic Susceptibility Test (blAST) is a population-level direct-sample ∼6-h phenotypic AST that measures protein synthesis through incorporation of non-canonical amino acids and click chemistry detection. This multi-site study evaluated blAST against urinary tract infection (UTI) pathogen isolates and clinical UTI samples. blAST agreement with reference broth microdilution (rBMD) was determined on 6 reference strains across 5 study sites and against 30 challenge strains. blAST was used to direct-test urine from clinically suspected UTIs (N = 740) and healthy donors (N = 12) for 9 antibacterials. Isolates from blAST-positive urine samples were tested by rBMD for comparative minimal inhibitory concentrations and susceptibility/resistance interpretations. Discordant isolates (N = 32) were retested by blAST. blAST demonstrated >92% between-site reproducibility on reference strains and >80% average agreement with rBMD for all tested antibacterials except nitrofurantoin (73%) on challenge strains. All urine samples from healthy donors were negative and 35% of UTI samples were blAST-positive. Concordance between methods was 62% with discordance predominantly blAST identifying resistance when rBMD determined susceptibility (35%). Discordant isolates retested by blAST agreed with rBMD (∼99%). blAST detected resistance in whole urine samples that was not found in individual isolates. Strong isolate agreement was found between blAST and rBMD methods on bacterial isolates. These data suggest that AST results from population-level direct-sample assays may not always agree with isolate-based protocols.