Off-panel resistance determinants in a high-resistance setting: comparative evaluation of syndromic multiplex PCR, metagenomic sequencing, and phenotypic AST in bloodstream infections

Rapid syndromic molecular panels substantially reduce turnaround time for bloodstream infection diagnosis by identifying common pathogens and selected antimicrobial resistance determinants directly from positive blood cultures. However, their fixed target repertoires may limit diagnostic coverage in settings with diverse endemic resistance mechanisms and clinically relevant off-panel organisms. This retrospective comparative diagnostic study evaluated direct-from-positive-blood-culture metagenomic next-generation sequencing (mNGS), the BioFire Blood Culture Identification 2 (BCID2) panel, and routine microbiological methods across 165 positive blood-culture episodes from 145 patients at a tertiary-care center with a high burden of antimicrobial resistance. Complete recovery of the routine-culture organism set was achieved in 155/165 episodes (93.9%) by mNGS and 142/165 (86.1%) by BCID2. Among 23 episodes with incomplete routine-set recovery by BCID2, mNGS restored complete coverage in 18 (78.3%). For meropenem resistance, sensitivity was numerically higher with mNGS than with BCID2 (90.0% versus 75.0%), whereas ceftriaxone-resistance sensitivity among Enterobacterales was numerically higher with BCID2 (97.5% versus 85.0%); neither paired difference was statistically significant. BCID2 meropenem very major errors were concentrated among non-fermenting gram-negative organisms, particularly Acinetobacter baumannii , and were frequently associated with off-panel bla OXA-23 detected by mNGS. Overall, targeted syndromic testing and mNGS demonstrated complementary diagnostic performance, with mNGS extending organism and resistance-mechanism coverage beyond predefined molecular targets, while phenotypic antimicrobial susceptibility testing remained necessary for definitive susceptibility assessment. IMPORTANCE Rapid molecular blood-culture panels provide clinically actionable results, but are inherently restricted to predefined pathogen and resistance targets. This limitation is particularly relevant in regions where local antimicrobial resistance epidemiology includes mechanisms not represented within commercial panel repertoires. In this study, metagenomic sequencing improved recovery of routine-culture organisms and identified off-panel carbapenem resistance determinants underlying several false-susceptible panel results, particularly OXA-23 in Acinetobacter baumannii . However, broader genomic detection did not consistently improve phenotypic resistance prediction across antimicrobial classes. These findings illustrate that diagnostic breadth and phenotypic predictive accuracy are distinct attributes and support the use of metagenomic sequencing as a complementary approach to rapid targeted testing and conventional antimicrobial susceptibility testing.