Comparison of the Performance of Phenotypic Methods for the Detection of Carbapenem-Resistant Enterobacteriaceae (CRE) in Clinical Practice

In order to investigate the diagnostic performance characteristics of four phenotypic assays in detecting carbapenem-resistant Enterobacteriaceae (CRE), we collected the CRE strains from infected patients. The results of carbapenemase gene detection, bla (KPC-2), bla (OXA-23), bla (NDM-1), bla (NDM-4), bla (NDM-5), bla (IMP-4), and bla (IMP-8), were used as a standard to evaluate the performances of combined disk test (CDT), modified carbapenem inactivation method(mCIM)/EDTA-modified carbapenem inactivation method(eCIM), NG-Test CARBA 5 (CARBA), and color developing immunoassay (CDI). The compliance of phenotype results based on CDT, mCIM/eCIM, CARBA, and CDI with genetic detection results was 94% (231/247), 95% (235/247), 98% (242/247), and 99% (246/247), respectively. CDT demonstrated a low specificity for carbapenemase detection, low negative predictive value (NPV), and low sensitivity for metallo-β-lactamase (79%, 55%, and 88%, respectively); it also failed to accurately detect IMP. The mCIM/eCIM assay had serious problems in detecting OXA-23-like carbapenemases. The sensitivity and specificity of CARBA and CDI were higher than those of the first two methods. However, CARBA did not cover the detection of OXA-23, while CDI cannot detect IMP-8, resulting in low NPVs (70% and 88%, respectively). In conclusion, CARBA and CDI assays are highly accurate except individual rare genes and allow direct genotype detections. CDT and mCIM/eCIM assays are moderately accurate and can only distinguish serine-β-lactamases from metallo-β-lactamases. Laboratories should choose the appropriate method that meets their needs based on its characteristic.