Abstract
BACKGROUND: The performance of multiplex PCR (mPCR) for detection of antimicrobial resistance from clinical isolates is unknown. We assessed the ability of mPCR to analyse resistance genes directly from clinical samples. Patients with orthopedic infections were prospectively included. Phenotypical and genotypical resistance was evaluated in clinical samples (synovial and sonication fluid) where identical pathogens were identified by culture and mPCR. RESULT: A total of 94 samples were analysed, including 60 sonication fluid and 34 synovial fluid samples. For coagulase-negative staphylococcus strains, mPCR detected resistance to oxacillin in 10 of 23 isolates (44%) and to rifampin in none of 6 isolates. For S. aureus isolates, detection rate of oxacillin and rifampin-resistance was 100% (2/2 and 1/1, respectively). Fluoroquinolone-resistance was confirmed by mPCR in all 3 isolates of Enterobacteriaceae, in enterococci resistance to aminoglycoside-high level was detected in 1 of 3 isolates (33%) and in streptococci resistance to macrolides/lincosamides in none of 2 isolates. The overall sensitivity for different pathogens and antimicrobials was 46% and specificity 95%, the median concordance was 80% (range, 57-100%). Full agreement was observed for oxacillin in S. aureus, vancomycin in enterococci, carbapenems/cephalosporins in Enterobacteriaceae and rifampin in Cutibacterium species. CONCLUSION: The overall sensitivity for detection of antimicrobial resistance by mPCR directly from clinical samples was low. False-negative mPCR results occurred mainly in coagulase-negative staphylococci, especially for oxacillin and rifampin. However, the specificity of mPCR was high and a positive result reliably predicted antimicrobial resistance. Including universal primers in the PCR test assay may improve the detection rate but requires additional sequencing step. TRIAL REGISTRATION: www.clinicaltrials.gov No. NCT02530229, registered at 21 August 2015 (retrospectively registered).