Fast-track identification of CTX-M-extended-spectrum-β-lactamase- and carbapenemase-producing Enterobacterales in bloodstream infections: implications on the likelihood of deduction of antibiotic susceptibility in emergency and internal medicine departments

Extended-spectrum-β-lactamase (ESBL)- and carbapenemase-producing Enterobacterales (EB) bloodstream infections (BSIs) are a relevant clinical issue worldwide, especially given their association with limited therapeutic options, poor outcomes, and growing community-onset prevalence [1‐5]. Since BSI patients are mainly managed in emergency and non-intensive wards [6, 7], rapid detection of ESBL- and carbapenemase-producing EB directly from positive blood cultures (BCs) in these settings is essential for early optimization of antibiotic therapy and improvement of clinical outcomes [8, 9]. Despite the complexity of the epidemiology, conjugative plasmids carrying CTX-M-type ESBL and KPC-, NDM-, VIM-, IMP-, and OXA-48-like genes were the main drivers of ESBL and carbapenemase dissemination in EB, respectively [1, 10, 11]. Recently, two lateral flow immunoassays (LFIAs), designed to detect CTX-M ESBL and the most widespread carbapenemases using specific monoclonal antibodies, have been optimally evaluated directly from positive EB BCs [12‐14]. Despite fast microbiology workflows on selected cohorts of patients being highly encouraged [8, 15], there is limited evidence on their implementation and firepower in lab routine [16]. This study aims at presenting a reliable fast-track diagnostics for the detection of CTX-M ESBL- (CTX-M-p) and carbapenemase-producers (CA-p) from positive EB BCs of patients admitted in emergency and internal medicine departments. In the carbapenem-sparing era, rapid identification of species and phenotype together with knowledge of local epidemiology in areas of high ESBL- and carbapenemase-producing EB endemicity may be crucial to indicate the likelihood of deduction of in vitro antibiotic susceptibility and drive antibiotic management on the same day of positive BC processing, at least 24 h before than conventional testing results.

Over the study period a total of 855 BCs obtained from patients suffering from a Gram-negative BSI were processed. Among these, 261 (30.5%) belonged to patients admitted in emergency and internal medicine departments, of which 90.4% (n=236) were reliably identified from bacterial pellet as EB species by MALDI-TOF MS analysis and included in the assessment of the fast-track workflow. Overall, polymicrobial BCs involving EB species were 12 (4.9%). Of these, ten BCs were not considered for study purpose due to unreliable MALDI-TOF identification score. The EB species mainly observed were Escherichia coli (68.2%; n=161), Klebsiella pneumoniae (18.2%; n=43), Enterobacter cloacae (4.2%; n=10), and Klebsiella aerogenes (3%; n=7).

The mean TtDP was 12.3 h (range 2–68). The mean TtR for conventional and fast-track workflow were 38 h (range 27–78) and 42 min (range 35–55), respectively.

Accuracy, sensitivity, specificity, PPV, and NPV of the fast-track BCs workflow for the detection of CTX-M-p and CA-p were shown in Table 1. Fast-track and conventional BC workflows results were discordant in one case regarding a CZA resistant KPC-producing K. pneumoniae that was misclassified as CTX-M-CA-np by the fast-track workflow.

Phenotypic characterization of EB species included in the study and likelihood of deduction of in vitro antibiotic susceptibility according to conventional culture-based results were reported in Table 2. Among E. coli isolates, 20.5% and 79.5% were CTX-M-p and CTX-M-CA-np, respectively. No CA-p E. coli were detected. CTX-M-p E. coli were in vitro susceptible to TZP (84.8%), C/T (97%), CZA (100%), and AK (75.8%). CTX-M-CA-np E. coli were largely in vitro susceptible to TZP (90.6%), CAZ (99.2%), FEP (100%), C/T (100%), CZA (100%), and AK (100%). Among AmpC/ESBL-np E. coli isolates, 9.5% (n=12) were resistant to TZP, whereas the only isolate resistant to CAZ was AmpC-p. Among K. pneumoniae isolates, 23.3%, 20.9%, and 55.8% were CTX-M-p, CA-p, and CTX-M-CA-np, respectively. CTX-M-p K. pneumoniae were in vitro susceptible to C/T (90%), CZA (100%), and AK (90%), whereas they showed susceptibility to TZP of 40%. All the CA-p K. pneumoniae isolates were pheno-genotypically characterized as KPC-producers, showing susceptibility to CZA and AK of 88.9% and 44.4%, respectively. Analysis of the KPC gene sequence of the only CZA-resistant CA-p K. pneumoniae isolate subsequently revealed to be a producer of KPC-31-carbapenemase harboring the mutation D179Y in the omega loop region of KPC-3 [23]. CTX-M-CA-np K. pneumoniae were in vitro susceptible to TZP (75%), CAZ (87.5%), FEP (100%), C/T (95.8%), CZA (100%), and AK (100%). Among AmpC/ESBL-np K. pneumoniae, 19.1% (n=4) were resistant to TZP, whereas among AmpC-p K. pneumoniae, 33% (n=1) and 66% (n=2) were susceptible to TZP and C/T, respectively. All the E. cloacae isolates were CTX-M-CA-np. They were in vitro susceptible to TZP (80%), CAZ (80%), FEP (80%), C/T (80%), CZA (100%), and AK (100%). All the AmpC/ESBL-p E. cloacae (n=2) were resistant to TZP, CAZ, FEP, and C/T. All the K. aerogenes isolates were CTX-M-CA-np. They were largely in vitro susceptible to FEP (85.7%), C/T (71.4%), CZA (100%), and AK (100%), whereas they showed poor susceptibility to TZP (28.6%) and CAZ (28.6%). Among AmpC-p K. aerogenes isolates, susceptibility to FEP and C/T were 100% (n=4) and 50% (n=2), respectively, whereas they were resistant to TZP and CAZ. The only AmpC/ESBL-p isolate showed resistance to TZP, CAZ and FEP, being susceptible to C/T. All the AmpC/ESBL-np isolates were susceptible to the antibiotics tested. All the Klebsiella oxytoca, Proteus mirabilis, Citrobacter spp, Pantoea agglomerans, Morganella morganii, and Salmonella spp isolates were CTX-M-CA-np, and they were susceptible to all the antibiotics tested. All the Serratia marcenscens isolates were CTX-M-CA-np, and they showed full susceptibility to FEP, C/T, and CZA. Susceptibility to TZP, CAZ, and AK was 50% since one of the isolates was phenotypically characterized as AmpC-p showing resistance to these antibiotics. The only Hafnia alvei isolate was classified as CTX-M-CA-np, being phenotypically characterized as AmpC-p, showing full susceptibility to TZP, FEP, CZA, AK, and resistance to CAZ, and C/T.

There was considerable agreement between the results provided by our fast-track workflow and conventional diagnostic methods. Despite the requirement for the later to confirm fast-track workflow results and to obtain a detailed pattern of drug activity, rapid tests for the detection of the more common resistance mechanisms in EB are an invaluable tool in setting timely antimicrobial stewardship interventions, thus potentially improving clinical outcomes, especially in BSI patients. Their implementation should prioritize patients at high risk for multidrug-resistant gram-negative bacteria infections to maximize cost-effectiveness [8]. In this 1-year study and considering disease prevalence, high PPV and NPV were found for our proposed fast-track workflow model.

The results provided were obtainable in less than 1 hour from the start of positive BC processing, which might favorably impact on antimicrobial stewardship interventions and be readily implemented into all laboratories with minimal training.

The fast-track BC workflow revealed to be particularly suited for the Italian and European epidemiology [1, 10], especially for cohorts of medical patients. In fact, EB were involved in almost all of BSI caused by gram-negative bacteria. E. coli and K. pneumoniae were the EB species mainly involved. They were found to be CTX-M-p in similar rates (20.5% and 23.3%, respectively), whereas carbapenemase production was observed only in K. pneumoniae isolates (20.9%). Similar epidemiological findings were found also by Fiori et al [16] that developed a timesaving and clinically relevant BC diagnostics based on the combined use of the MALDI BioTyper and a qualitative genotypic test to detect CTX-M-p and main CA-p in patients suffering from E. coli and K. pneumoniae BSI.

Concerning CA-p detection, since a CZA-resistant KPC-producing K. pneumoniae isolate was misclassified as CTX-M-CA-np by the fast-track diagnostics, a relevant issue emerged in this study [23]. In fact, KPC variants harboring mutations in omega loop, phenotypically characterized by impaired carbapenemase activity, restoration of carbapenems susceptibility, and expression of ESBL phenotype [24], can be undetected by LFIAs. Therefore, in patients with history of recent CZA treatment, a negative phenotypic result for KPC production should be confirmed by molecular assays.

Rapid detection of CTX-M-p, CA-p, or CTX-M-CA-np may also allow the prediction of multidrug resistance phenotypes and in vitro susceptibility based on local epidemiology of the EB species involved. Indeed, detection of CTX-M-p allows the early prediction of in vitro susceptibility to cephalosporins/beta-lactamase inhibitor combinations, such as C/T and CZA. TZP also showed a satisfactory in vitro activity against CTX-M-p E. coli, being its use considered according to BSI severity and source of infection [22]. Detection of KPC-producers suggests the probable efficacy of CZA, even though isolation of CZA-resistant KPC producing-EB could occur especially following CZA treatment. Detection of CTX-M-CA-np predicts in vitro susceptibility to CAZ among E. coli and K. pneumoniae isolates, but the presence of AmpC-p should be considered and switching to FEP always pondered. Of note, among CTX-M-CA-np E. coli and K. pneumoniae, 10.1% (n=15) that were classified as AmpC/ESBL-np showed in vitro TZP resistance and third-generation cephalosporin susceptibility. These strains were considered an emerging cause of BSI [25] and were the target of phenotypic and genotypic characterization [26] that did not fully elucidate the underlying resistance mechanism, even though they showed TZP activity on a murine model. Due to limited evidence on its generalizability to clinical practice, further studies are needed before considering TZP as a valid option against these strains in BSI patients. Among all the other than E. coli and K. pneumoniae EB, deduction of in vitro susceptibility should be based on EB species since the fast-track workflow classified all the isolates as CTX-M-CA-np. Therefore, empirical antibiotic therapy should consider the presence of inducible and derepressed AmpC- or AmpC/ESBL-p species (Enterobacter spp, Citrobacter freundii, Serratia marcescens, Morganella morganii, Hafnia alvei, and Providencia spp) [27] and avoid third generation cephalosporins [28] as well as TZP. These observations are limited due to the small sample size obtained in this study for these agents.

Strengths of this study include its pragmatic design and potential contribution to active programs of antimicrobial stewardship focused on selected cohorts of patients to maximize cost-effectiveness.

This study has some limitations. The fast-track BC workflow presented largely depends by a reliable direct MALDI-TOF identification and was designed according to the epidemiology of our hospital’s emergency and internal medicine departments and so it limits its generalizability to other settings. EB strains expressing carbapenemases other than those of the KPC type were not evaluated (i.e., blaVIM, blaIMP, blaOXA-48, blaNDM). Moreover, our fast-track BC workflow probably needs to be reviewed to maximize cost-effectiveness. Performing detection of CTX-M ESBL on both E. coli and K. pneumoniae species and detection of carbapenemases only on K. pneumoniae isolated in patients with history of recent hospitalization seems to be advisable. However, implementing fast-track diagnostics in lab routine means to be prompt to change it according to epidemiology and/or clinical reasons.

In conclusion, association of fast-track diagnostics for the detection of CTX-M-p and CA-p in BSI patients to conventional culture-based antimicrobial susceptibility testing in areas of high ESBL- and carbapenemase-producing EB endemicity may predict the likelihood of deduction of in vitro antibiotic susceptibility for each EB species on the same day of positive BC processing. Further studies are warranted to assess the impact of the implementation of fast-track BC workflow in antimicrobial stewardship programs on patient’s outcome and determining its cost effectiveness.