In our retrospective cohort of ICU patients, 16% of the strains of E. coli isolated from clinically relevant specimens were resistant or intermediate to PIP-TAZ. This is far more than the rates observed by the French National Observatory for Epidemiology of Bacterial Resistance to Antimicrobials (ONERBA) that reports a 4% rate of PIP-TAZ intermediate or resistant strains among the strains of E. coli isolated in hospitalised patients . In ICU, the PIP-TAZ resistance rate of E. coli is much higher. Our results confirmed those reported by Mohammedi et al. who noted an increase in the isolation rate of E. coli intermediate or resistant to PIP-TAZ in ICU patients from 6% in 1995 to 17% in 1999 . But Mohammedi et al. did not characterise the mechanisms of resistance to PIP-TAZ. In our cohort, two among 13 PIP-TAZ resistant strains produced extended-spectrum β-lactamase. The production of high-level penicillinase was the predominant mechanism of resistance, identified in 11/13 PIP-TAZ resistant strains. This may have important consequences for the empirical choice of antimicrobial agents: these strains retained susceptibility to cefotaxime. The 2% (2/83) prevalence of ESBL strains in our cohort was similar to the prevalence recently described in French hospitals . Recently, several studies revealed emergence of new ESBL-producing E. coli especially within the community, the CTX-M-producing isolates . A limitation of our study is that underlying mechanisms of resistance were determined only using interpretative reading of the antibiogram. We did not performe any molecular study to identified the different β-lactamases.
In our cohort, a prolonged hospital stay prior to the diagnosis of E. coli infection or a recent antibiotic therapy containing amoxicillin or amoxicillin-acid clavulanic favoured infections with a PIP-TAZ resistant strain of E. coli. Mohammedi et al. conducted a multivariate analysis and identified prior use of amoxicillin and amoxicillin-acid clavulanic as a major risk factor of infection with PIP-TAZ resistant or intermediate strains of E. coli . Their results were in accordance with previously published data establishing a link between prior antibiotherapy and infection with penicillin resistant strains [24, 25]. In our study, as in any observational study, the missing data could have reduce the strength of our observations. The potential for unmeasured confounding data must be recognized.
In our cohort, we did not find any impact of the PIP-TAZ resistance phenotype on the prognosis of infections with E. coli. Several previous studies demonstrated that some patterns of resistance were significantly associated with mortality of E. coli infections [11, 12]. Our study is the first analyzing the association of PIP-TAZ resistance with the outcome of E. coli infections. One explanation to the absence of prognosis impact of PIP-TAZ resistance in our cohort is that PIP-TAZ resistance did not result in a decrease of the proportion of adequate initial antimicrobial therapy. Monocentric nature of our study could explain this result, since PIP-TAZ was only rarely prescribed empirically in our ICU. In our cohort, predominance of medical admissions leading to a low proportion of intra-abdominal infections could have contributed to reduced prescription of PIP-TAZ. Urinary tract infections and pneumonia were the most frequent reported infections in our patients. PIP-TAZ is recommended for severe community-acquired intra-abdominal infections and for health care-associated intra-abdominal infections . But, in case of pneumonia, it is recommended only for hospital-acquired pneumonia, ventilator-associated pneumonia, and health care-associated pneumonia in patients with late-onset disease or risk factors for multidrug-resistant pathogens . Third generation cephalosporins are the reference treatment for complicated community-acquired urinary tract infections and is recommended for nosocomial urinary tract infections [26, 27]. Another explanation to the absence of prognosis impact of PIP-TAZ resistance in our cohort is that our study was likely under powered to demonstrate a clinically important difference.
Our findings could have an impact on the choice of the empirical antibiotic prescription in case of Gram negative bacilli infection in critically ill patients. The selection of appropriate initial antibiotic therapy remains a challenge for the physician, who must balance the need for eradication of infection against the selection of resistant pathogens. The prognostic importance of appropriate initial antimicrobial therapy both in timing and efficacy on the causal micro-organism(s) has been underlined in many studies [28, 29]. Prescription of PIP-TAZ in case of Gram negative bacilli infection in ICU is justified by its ability to treat both infections with enterobacteriaceae and with non-fermentative Gram-negative bacilli such as Pseudomonas aeruginosa, which are particularly frequent in ICU patients. The SOAP study conducted in European intensive care units revealed that the most common organisms isolated in the course of sepsis were Staphylococcus aureus (30%), Pseudomonas species (14%), and Escherichia coli (13%). The broad spectrum of activity of PIP-TAZ led some authors to consider it as the standard therapy for many infections such as ventilator-associated pneumonia and abdominal infections [30, 31]. The high isolation rate of E. coli intermediate or resistant to PIP-TAZ that we report in ICU patients exposes to failure of empirical treatment with PIP-TAZ. Nevertheless our study does not allow to draw firm conclusions on the best empirical antibiotic therapy in case of E. coli infections in ICU. Finally, it highlights the importance of the local surveillance of the epidemiology of microorganisms in devising antibiotic strategies in a specific ICU.