Abstract
Stenotrophomonas maltophilia causes high-mortality infections in immunocompromised hosts with limited therapeutic options. Many U.S. laboratories rely on commercial automated antimicrobial susceptibility tests (cASTs) and use CLSI breakpoints (BPs) for S. maltophilia. However, contemporary data on these systems are lacking. We assessed performance of Vitek 2, MicroScan WalkAway, and Phoenix relative to that of reference broth microdilution for trimethoprim-sulfamethoxazole (SXT), levofloxacin (LEV), minocycline (MIN), and ceftazidime (CAZ) with 109 S. maltophilia bloodstream isolates. Using CLSI breakpoints, categorical agreement (CA) was below 90% on all systems and drugs, with the exception of SXT by MicroScan (98.1%) and Phoenix (98.1%) and MIN by MicroScan (100%) and Phoenix (99.1%). For SXT, Vitek 2 yielded a 77.1% CA. LEV and CAZ CA ranged from 67% to 85%. Very major errors (VME) were >3% for SXT (MicroScan, Phoenix), LEV (MicroScan), and CAZ (all systems). Major errors (ME) were >3% for SXT (Vitek 2), LEV (Phoenix), and CAZ (MicroScan, Phoenix). Minor errors were >10% for CAZ and LEV on all systems. Data were analyzed with EUCAST pharmacokinetic/pharmacodynamic CAZ, LEV, ciprofloxacin (CIP), and tigecycline (TGC) breakpoints when possible. CA was <90% for all. VME were >3% for CAZ (all systems), LEV (MicroScan), and TGC (Vitek 2), and ME were >3% for LEV (MicroScan), CAZ (all systems), ciprofloxacin (Vitek 2 and MicroScan), and TGC (Vitek 2, Phoenix). Minor errors (MI) were >10% for all agents and systems, by EUCAST breakpoints with an intermediate category (LEV, CAZ, CIP). Laboratories should use caution with cASTs for S. maltophilia, as a high rate of errors may be observed.