Discussion

In this study, CTX-M detection was 90% sensitive and 100% specific for predicting ceftriaxone resistance in Enterobacterales. VCUHS 2023 antibiogram data indicate ceftriaxone susceptibility for 90.4% of non-inducible AmpC-producing Enterobacterales. Our ASP recommends ceftriaxone for most Enterobacterales bacteremia without BCID2-identified resistance genes. However, providers are often reluctant to de-escalate without confirmed susceptibility, as there are multiple potential mechanisms of gram-negative antimicrobial resistance. Production of a β-lactamase is the most common in Enterobacterales, but nearly 3,000 unique enzymes exist. Although CTX-M accounts for >90% of ESBL-producing organisms in the United States, there are other enzymes BCID2 cannot detect that cause ceftriaxone resistance. ^{Reference Castanheira, Simner and Bradford8} Interestingly, in this study, CTX-M negative isolates were more often susceptible to ceftriaxone than piperacillin/tazobactam, which is commonly used as empiric gram-negative therapy at VCUHS. Approximately 10% of isolates tested piperacillin/tazobactam S-DD, which requires treatment with high-dose extended infusion to achieve adequate time above the minimum inhibitory concentration (MIC). This pattern may reflect selective pressure due to piperacillin/tazobactam overuse and increases the risk for inadvertent selection of inappropriate empiric antibiotics.

We add to the body of literature on the use of multiplex PCR in management of gram-negative bacteremia. A retrospective, observational study in India (n = 200; 183 of 200 gram-negative isolates) reported 97% concordance between BCID2 resistance gene detection and phenotypic drug susceptibility. The 6 discordant isolates were in polymicrobial cultures. ^{Reference Vineeth, Nambi and Gopalakrishnan9} A prospective, single-center study in Korea (n = 117; 64 of 117 gram-negative isolates) found BCID2 accurately predicted phenotypic susceptibility for 96.9% of gram-negative isolates. Two strains of ESBL Serratia marcescens were not identified as CTX-M positive. ^{Reference Kim, Yun, Cho, Lee and Nam10} Another prospective, single-center study in Germany (n = 180; 93 of 180 gram-negative isolates) found 16 gram-negative isolates resistant to third-generation cephalosporins, but only 12 identified as CTX-M positive by BCID2. Three of the CTX-M negative isolates carried bla _SHV and/or bla _TEM. The resistance mechanism of the fourth isolate was not determined. ^{Reference Berinson, Both and Berneking11} Finally, a prospective, multicenter study in France (n = 152; 98 of 152 gram-negative isolates) found 2 false-positive CTX-M and 2 ESBL isolates of E. cloacae not identified as CTX-M positive by BCID2. ^{Reference Caméléna, Péan de Ponfilly and Pailhoriès12}

Our study has several limitations. Some Enterobacterales, such as Citrobacter freundii, Hafnia alvei, and Yersinia enterocolitica, are not identified at the genus or species level by BCID2. Due to the risk of inducible AmpC resistance, ceftriaxone should be avoided in these isolates. Furthermore, BCID2 does not identify constitutive AmpC production. The 3 ceftriaxone-resistant isolates in this study remained susceptible to cefepime and meropenem but were variably susceptible to piperacillin/tazobactam, a phenotype possibly suggestive of AmpC production. Finally, these data may not be generalized to institutions with differing patient populations or antibiograms, such as locations with higher prevalence of non-CTX-M ESBLs.

Treatment of sepsis using antibiotics with local susceptibility ≥90% results in decreased in-hospital mortality. This threshold is often used as a benchmark for selection of appropriate empiric therapy. ^{Reference Chang, Hsieh, Yang, How, Chen and Meng13} In our study, ceftriaxone demonstrated 98.8% susceptibility for Enterobacterales negative for resistance genes on BCID2. These data suggest that early de-escalation to ceftriaxone in patients with Enterobacterales bacteremia without resistance genes identified by BCID2 may be appropriate. Providers must still assess patients individually for severity of infection, suspected source, and risk for polymicrobial infection. In patients at high risk for drug-resistant pathogens, infectious diseases and/or ASP-trained individuals should be involved in the interpretation of PCR results and subsequent treatment recommendations. We highlight an opportunity for safe and effective early de-escalation of antibiotics for the treatment of Enterobacterales bacteremia in appropriate patient populations.