Early use of imipenem/cilastatin and vancomycin followed by de-escalation versus conventional antimicrobials without de-escalation for patients with hospital-acquired pneumonia in a medical ICU: a randomized clinical trial

Early use of broad-spectrum antimicrobials and subsequent de-escalation (DE) after microbiologic culture results may minimize the emergence of drug-resistant organisms [1‐3] and costs [1] during the treatment of patients with hospital-acquired pneumonia (HAP). This strategy may reduce the overall duration of antimicrobial treatment [4] and mortality rates [5, 6]. These studies, however, were observational, with successful DE indicating less-resistant pathogens and/or less severity of infection.

The combination of carbapenem and vancomycin is frequently prescribed for the treatment of HAP, in which the prevalence of infection by multidrug-resistant (MDR) organisms is high. Although inadequate initial therapy may increase mortality rates [7], the frequent use of carbapenem plus vancomycin predisposes to the emergence of MDR organisms. Vancomycin may select for resistant pathogens once colonization takes place [8], and vancomycin use has been associated with increases in vancomycin-resistant enterococci (VRE) [9] and vancomycin-intermediate and -resistant S. aureus [10]. Treatment with carbapenems has also been reported to increase the overall risk for emergence of MDR organisms [11‐13]. Moreover, the efficacy of vancomycin as empiric therapy for HAP in hospitals with a high prevalence of methicillin-resistant Staphylococcus aureus (MRSA) has not been completely addressed. Vancomycin is a relatively large molecule, with poor penetration into the alveolar lining and alveolar macrophages [14, 15]. In addition, the cure rate for vancomycin in patients with MRSA pneumonia has been disappointing [16, 17]. A study of febrile neutropenic patients with MRSA infection showed that these patients did not experience worse outcomes when vancomycin therapy was delayed until definitive diagnosis [18]. The use of carbapenem, which has the broadest spectrum of activity against Gram-negative organisms, must also be controlled to limit the selection and spread of MDR pathogens.

To examine these concerns, we conducted a randomized clinical trial in a large tertiary care center in Korea, where the methicillin resistance rate of S. aureus isolated from blood has been reported to be 72% [19]. In this trial, we compared the effects of early treatment with broad-spectrum antimicrobials followed by subsequent DE, with conventional antimicrobial regimens.

This prospective, randomized clinical trial showed that the adequacy of initial antimicrobial therapy is significantly higher with early use of broad-spectrum antimicrobials followed by subsequent DE than with it is conventional therapy. The higher antimicrobial adequacy observed in the DE group was mainly due to MRSA coverage by early administration of vancomycin. However, no significant between-group differences were noted in overall duration of antimicrobial therapy and mortality rates, with the DE group showing a higher rate of MRSA acquisition. Although vancomycin was administered early, the mortality associated with MRSA infections was higher among DE patients. Thus, early administration of vancomycin failed to show therapeutic benefits in our study, and early MRSA coverage before microbiologic documentation seems to be unnecessary.

Early use of broad-spectrum antimicrobials with subsequent de-escalation is considered a promising strategy for optimizing the responsible use of antimicrobials [1]. An observational study of 398 ICU patients with suspected VAP reported that the mortality rate was significantly (P = 0.001) lower in patients with DE (17%) than in those with no change in therapy (23.7%) or escalation (42.6%) [5]. That study, however, was observational, with no randomization, and other factors, such as baseline disease severity, may have influenced treatment outcomes, rather than the DE itself.

Previous studies, however, have reported that DE was associated with higher antimicrobial adequacy and more favorable outcomes [6, 25‐27]. The difference between these results and ours may be attributable to various factors. First, our strategy may have improved antimicrobial adequacy for Gram-positive organisms, most of which were MRSA. Despite increased antimicrobial adequacy of vancomycin against Gram-positive organisms in the DE group, vancomycin had only limited overall efficacy in the treatment of HAP, with no difference in treatment outcomes. Moreover, the mortality rate was higher in our DE than in our NDE group. Vancomycin has been the only antimicrobial available for the treatment of MRSA pneumonia for years [14], but its cure rate has been disappointing [16, 17]. Also, despite the broad antimicrobial spectrum of imipenem, its adequacy for Gram-negative organisms was similar in our DE and NDE groups (64.3% versus 85.7%; P = 0.19). In a previous study, the adequacy of treatment for Gram-negative organisms was found to be higher in patients treated with a DE regimen, in which the major comparison was between cefotaxime and imipenem, which were used in 39% and 100% of NDE and DE patients, respectively [6]. In contrast, piperacillin/tazobactam (87.3%) and ciprofloxacin (61.8%) were the most frequently used initial antimicrobials in our NDE group. Previous studies have suggested that piperacillin/tazobactam and imipenem/cilastatin have similar efficacy and safety profiles [28‐30]. Only one patient had ESBL-producing Enterobacteriaceae, and broad-spectrum antimicrobials may have different sensitivity patterns at other centers. In addition, the definition of "de-escalation" differed among studies, in that most defined DE with regard to Gram-negative organisms. For example, the activity spectrum of antimicrobials was ranked against Gram-negative bacteria, with DE defined as a switch to or discontinuation of a drug class with a less-broad spectrum of coverage [5]. Others have defined de-escalation similarly, by ranking the antimicrobial spectra of these agents [6, 31]. In contrast, imipenem/cilastatin and vancomycin were de-escalated individually, with the former de-escalated only in patients with Gram-negative organisms. Categorizing all these definitions by using the single term "de-escalation" is confusing and may result in divergent findings.

The strengths of our study include its prospective design, high initial prevalence of MDR organisms, and relatively high severity of infection. To our knowledge, this is the first randomized clinical trial comparing the DE of imipenem plus vancomycin with that of non-carbapenem, non-vancomycin in a medical ICU. This study, however, had several limitations. First, it was performed in a single center MICU, which may have biased patient selection. The proportion of patients with VAP was relatively small, and carbapenems other than imipenem, such as doripenem, may yield different results. Therefore, the generalizability of this study may be limited. Furthermore, the baseline prevalence of MDR organisms was higher in the DE than in the NDE group, which may have influenced treatment outcomes [32, 33]. However, we sought to minimize this effect by excluding these patients from subsequent analyses for emergence of MDR organisms. Moreover, this study was not double-blinded. Indeed, the attending physicians showed a tendency not to de-escalate. Nevertheless, the DE rates were relatively high (83.3% for vancomycin and 84.8% for imipenem/cilastatin). Another limitation was our inability to identify pathogens in 50% of our study patients. The percentage of patients who underwent bronchoalveolar lavage at the time of enrollment was relatively low; thus, our rate of obtaining adequate specimens was not high. A possibility of cross-contamination exists as a source of MDR organisms. We also used the Clinical and Laboratory Standards Institute (CLSI) standards, rather than the European Committee on Antimicrobial Susceptibility Testing (EUCAST) standards to define antimicrobial susceptibility, which may have resulted in a small difference in the incidence of MDR organisms. In addition, our secondary end points, including mortality rates and emergence of MDR organisms, may not have been sufficiently powered to detect significant differences. Also, rectal swabs to identify carriers were not routinely checked. Finally, the dosage of imipenem (2 g/day) may have been insufficient for patients with Pseudomonas pneumonia.

Although early use of broad-spectrum antimicrobials significantly increased initial antimicrobial adequacy, treatment outcomes were not improved. The therapeutic advantage with early use of broad-spectrum antimicrobials, especially vancomycin, was not evident in this study. Our results should be considered as hypothesis generating, as discussed earlier, and additional clinical research to verify our observations is needed.

♦ Although early use of broad-spectrum antimicrobial agents significantly increased the initial adequacy against multidrug-resistant pathogens among critically ill patients with hospital-acquired pneumonia, it failed to improve survival.

♦ Our data suggest that efficacy of early vancomycin administration is limited, considering that emergence of MRSA in the de-escalation group was even higher than that in the non-de-escalation group.