Abstract
Multilocus sequence typing (MLST) is one of the most reliable methods for typing of Escherichia coli, including extended-spectrum-β-lactamase-producing E. coli (ESBL-EC). We investigated the performance of a new typing method, SpectraCell RA (River Diagnostics, Madison, WI), in comparison on MLST on a well-defined collection of ESBL-EC isolates obtained from chicken meat and humans. Ninety-two ESBL-EC isolates obtained from meat and 59 ESBL-EC isolates obtained from human rectal swabs and clinical blood cultures were typed using MLST and SpectraCell RA. The sensitivity and specificity of SpectraCell RA were calculated, using MLST as a reference method. Subsequently, the results of SpectraCell RA were used to determine the relatedness of ESBL-EC isolates from chicken and humans. Using MLST as the gold standard, the performance of SpectraCell RA was evaluated for 3 different cutoff values: 0.99975, 0.99955, and 0.99935. Depending on the cutoff value, the sensitivity was mediocre to unacceptably low, with values of 9.4%, 43.9%, and 66.7%, respectively. When sensitivity increased, the specificity decreased rapidly, from 95.6% to 69.8% and 34.4%, respectively. Also, the number of clusters containing both human and meat samples varied from 0 (0.0%) to 14 (38.9%). Our study shows that SpectraCell RA is not a suitable typing method for ESBL-EC when evaluating relationships of ESBL-EC at the population level.