The present study is consistent with previously published data where CR-PA is a growing issue in both the ICU and non-ICU settings. Indeed, 59% of the CR-PA in the present study were isolated from patients in non-ICU settings consistent with the findings that CR-PA is not only an ICU based pathogen [2]. Similarly, respiratory tract was the most common culture source in both the ICU and non-ICU patients. These data provide significant insights for therapeutic options for hospital acquired pneumonia (HAP) including hospital acquired pneumonia requiring ventilation (V-HAP) and HAP not requiring ventilation (NV-HAP) for the non-ICU isolates as these patients may undergo transitions from the ward to the ICU for escalated care. Data from ICU patients may provide insights into the susceptibility pattern of CR-PA responsible for ventilator associated pneumonia (VAP) [6].
V-HAP represents a challenging clinical syndrome where mortality is estimated at 28%, notably higher than VAP (18%) and NV-HAP (15%) [6]. The mechanisms behind increased mortality in V-HAP compared with the syndromes are unclear, however; inappropriate empiric therapy due to higher prevalence of resistant organisms has been associated with increased complications [7]. Cefepime represents first line therapy for HAP/VAP [5]. However, regardless of ICU or non-ICU status in the present study, cefepime activity was markedly worse than ceftolozane/tazobactam or ceftazidime/avibactam in the setting of CR-PA. This is echoed in a retrospective cohort study where infection with CR-PA that were susceptible to other typical β-lactam agents (i.e., cefepime, ceftazidime, and piperacillin/tazobactam) was associated with higher 30-day mortality than infection with CR-PA resistant to these agents [8]. Although treatment regimens were not assessed, a potential explanation for this finding is the use of non-pharmacodynamically optimized dosing regimens in the setting of higher MICs to the susceptible β-lactams due to cross-resistance. We previously found that the elevated MICs to cefepime and ceftazidime in CR-PA, although susceptible per interpretive criteria, necessitate pharmacodynamically optimized doses to achieve PKPD targets [9]. Further investigation is needed to evaluate if pharmacodynamically optimized dosing can improve outcomes for CR-PA susceptible to typical β-lactams. When treating CR-PA, clinicians should consider more potent alternatives (e.g., ceftolozane/tazobactam or ceftazidime/avibactam) until such data are available.
Cross-resistance amongst P. aeruginosa must also be considered during transitions of care from non-ICU to ICU settings which may be prominent in V-HAP [5, 6]. Meropenem is often considered an escalation in this setting of cefepime failure. The present cohort exclusively comprised of carbapenem-resistant isolates however Lob and colleagues reported in a US cohort of P. aeruginosa that cefepime-non-susceptibility was accompanied with meropenem-susceptibly in only 36% and 40% of isolates in the ICU and non-ICU compared with 77% and 84% of isolates for ceftolozane/tazobactam in each setting, respectively [10]. Similar to our findings, the same authors found meropenem-non-susceptibility was associated with 46% and 49% susceptibility to cefepime in the ICU and non-ICU, respectively. These data reinforce that P. aeruginosa with reduced susceptibility to cefepime and/or carbapenems may carry cross-resistance to the other agent. Contemporary antibiograms do not account for the fact that resistance to one β-lactam agent may be accompanied with elevated MICs to another. Other interventions (e.g., MDR- antibiograms or patient risk based algorithms) to stratify patients who would benefit from empiric escalation to more potent agents (e.g., ceftolozane/tazobactam and ceftazidime/avibactam) are warranted.
Cascade reporting of susceptibility results have been a standard practice for antimicrobial stewardship programs. In this model, expanded agents (e.g., ceftolozane/tazobactam and ceftazidime/avibactam) are tested when specific criteria are met (e.g., carbapenem-non-susceptibility). Of note, novel agents may not be immediately available on automated susceptibility testing platforms and clinical laboratories must utilize other methods for testing such as gradient diffusion strips or disk diffusion which may take another 24 h for results [11]. A quasi-experimental study in a regional health system found that the median time from known meropenem-non-susceptibility to ceftolozane/tazobactam testing results was 25.4 h [12]. Based on the present study, knowing the carbapenemase status via phenotypic or genotypic methods may increase the reliability to appropriately select antimicrobials in the setting of CR-PA while awaiting confirmatory susceptibility testing. Indeed, the mCIM was used in this study which requires 18–24 h to obtain a result, which is likely similar to the timeline for confirmatory susceptibility testing [4, 11, 13]. However, more rapid methods (i.e., CarbaNP, immunoassays) or molecular diagnostics (i.e., the CarbaR) are available in as little as 30 min which would provide actionable data sooner [4, 10, 13]. Figure 3 describes a potential therapeutic decision making algorithm in response to carbapenemase testing as described in Fig. 2. Indeed, ruling out a carbapenemase (phenotypically or genotypically) can provide clinicians confidence in starting ceftolozane/tazobactam or ceftazidime/avibactam therapy while awaiting confirmatory susceptibility testing. Similarly, Molecular detection of KPC or GES can alert clinicians to select ceftazidime/avibactam as it is more likely to be active. Finally, detection of metallo-beta-lactamases results in poor activity of either ceftolozane/tazobactam or ceftazidime/avibactam alerting clinicians to use alternative agents or combinations. Our data shows such carbapenemase status therapy stratification can be useful in both the ICU and non-ICU setting.
Interestingly, phenotypic and genotypic mis-matches were detected in our cohorts (i.e., mCIM-negative, GES-positive). Indeed, not all GES-subtypes have been considered carbapenemases some are considered ESBLs [14, 15]. Although PCR-testing used cannot differentiate between alleles previously categorized as ESBLs or carbapenemases, detection of such enzymes can still guide therapeutic decisions as ceftazidime/avibactam is more likely to be active than ceftolozane/tazobactam regardless of variant (Fig. 2) [3]. More data are needed to dictate optimal therapy when GES-harboring P. aeruginosa are detected.
