Comparative Analysis of Bacillus cereus Group Isolates' Resistance Using Disk Diffusion and Broth Microdilution and the Correlation between Antimicrobial Resistance Phenotypes and Genotypes

Bacillus cereus group isolates (n = 85) were screened for phenotypic resistance to 18 antibiotics using broth microdilution and CLSI M45 Bacillus spp. breakpoints. The susceptibility to 9 out of 18 antibiotics was also tested using disk diffusion method and M100 Staphylococcus spp. breakpoints when available. Overall, a high prevalence of susceptibility to clinically relevant antibiotics was identified using broth microdilution. For most tested antibiotics, a poor correlation was found between zones of inhibition and MICs. Using the broth microdilution results as a reference for comparison, we identified high error rates and low categorical agreement between results produced using disk diffusion and broth microdilution for the seven tested antibiotics with defined breakpoints. This suggests that disk diffusion should be avoided for AST of B. cereus group isolates. Further, we detected antimicrobial resistance genes with ARIBA and ABRIcate to calculate the sensitivity and specificity for predicting phenotypic resistance determined using broth microdilution based on the presence of detected antimicrobial resistance genes (ARGs). ARGs with poor sensitivity and high specificity included rph (rifampicin, 0%, 93%), mph (erythromycin, 0%, 99%), bla1 (penicillin, 29%, 100%), and blaZ (penicillin, 56%, 100%). Compared to penicillin, bla1 and blaZ had lower specificity for the prediction of ampicillin resistance. Overall, none of the ARGs had both high sensitivity and specificity, suggesting the need for further study of the mechanisms underlying phenotypic antimicrobial resistance in the B. cereus group. IMPORTANCE Bacillus cereus group includes human pathogens that can cause severe infections requiring antibiotic treatment. Screening of environmental and food isolates for antimicrobial resistance can provide insight into what antibiotics may be more effective therapeutic options based on the lower prevalence of resistance. Currently, the comparison of antimicrobial susceptibility testing results using the disk diffusion method is complicated by the fact that many previous studies have used Staphylococcus spp. breakpoints to interpret their results. In this study, we compared the results of disk diffusion interpreted using the Staphylococcus spp. breakpoints against the results of broth microdilution interpreted using Bacillus spp. breakpoints. We demonstrated that the disk diffusion method does not produce reliable results for B. cereus group isolates and should therefore be avoided. This study also provides new insight into poor associations between the presence of antimicrobial resistance genes and resistance phenotypes for the B. cereus group.