LAUSR

Purpose To evaluate the performance of five widespread commercial products for colistin and polymyxin B susceptibility testing in China for mcr-positive and -negative Escherichia coli and Klebsiella pneumoniae. Methods A total of 132 E. coli and 83 K. pneumoniae strains (including 68 mcr-1-positive E. coli and 28 mcr-8-positive K. pneumoniae) were collected. We analysed the performance of colistin susceptibility (with Vitek 2 and Phoenix M50) and the performance of polymyxin B susceptibility (with DL-96II, MA120, and a Polymyxin B Susceptibility Test strip; POL E-strip). Broth microdilution was used as the gold standard. Categorical agreement (CA), essential agreement (EA), major error (ME), and very major error (VME) were calculated for comparisons. Results For E. coli, the total CA, EA, ME, and VME to colistin were as follows: Vitek 2, 98.5%/98.5%/0%/2.9%; and Phoenix M50, 98.5%/97.7%/0%/2.9%. The total CA, EA, ME, and VME to polymyxin B were as follows: POL E-strip, 99.2%/63.6%/1.6%/0%; MA120, 70.0%/-/0%/58.8%; and DL-96II, 80.2%/-/1.6%/36.8%. Only Vitek 2 and Phoenix M50 presented satisfactory performances for mcr-1-positive E. coli. For K. pneumoniae, the total CA, EA, ME, and VME to colistin were as follows: Vitek 2, 73.2%/72.0%/0%/61.6%; and Phoenix M50, 74.7%/74.7%/0%/58.3%. The total CA, EA, ME, and VME to polymyxin B were as follows: POL E-strip, 91.6%/74.7%/2.1%/16.7%; MA120, 92.8%/-/2.1%/13.9%; and DL-96II, 92.2%/-/2.1%/8.3%. All systems were unsatisfactory for mcr-8-positive K. pneumoniae. When the susceptibility of mcr-negative strains was tested, all systems presented excellent performance. Conclusion Vitek 2 and Phoenix M50 with colistin for E. coli showed acceptable performance regardless of mcr-1 expression, while DL-96II, MA120, and the POL E-strip performed worse for mcr-1-positive strains. Furthermore, mcr-8 greatly affected the performance of all systems with both colistin and polymyxin B for K. pneumoniae isolates.