Emergence of multidrug-resistant Gram-negative bacteria during selective decontamination of the digestive tract on an intensive care unit—authors' response

Abstract

Sir, We thank Abecasis and colleagues for their comments on our paper on the detection of extended-spectrum b-lactamase (ESBL)-producing positive aerobic Gram-negative bacilli (ESBL-GNB) in patients on the intensive care unit (ICU) of our hospital, where selective decontamination of the digestive tract (SDD) is routinely used in mechanically ventilated patients. Their main comment is that the link that we suggest between SDD use and the detection of these strains is only circumstantial; they also argue that the occurrences of such isolates is related to their resistance to tobramycin, an antibiotic in the SDD regimen which could be replaced by neomycin or paromomycin. We agree that during treatment with SDD, ESBL-GNB isolates will only reach detectable numbers in cultures from patients if they have decreased susceptibility to tobramycin. We isolated four genetically different strains that carried identical plasmids, had identical resistance patterns, and carried identical ESBL genes. As we noted in our paper, the emergence of the four strains could be due to: (i) an increase in these strains from below to above detection level; (ii) acquisition of the strains carrying the resistance plasmid; or (iii) acquisition of the resistance plasmid by strains that belong to the normal flora of the patients. We do not state that the emergence of these strains would not have occurred if these patients had not received SDD; we only want to show that SDD does not prevent the emergence of such strains and possibly, like any other antimicrobial treatment including cephalosporins, even promotes their emergence. Possibility (i) is unlikely, as it implies that isolates with decreased cefotaxime and tobramycin susceptibility are frequently missed during screening in an SDD setting. Acquisition of strains with identical plasmids seems improbable because this presumes a widespread presence of plasmids with identical RFLP patterns in different isolates. RFLP analyses of plasmids isolated from two ESBL-positive and one ESBL-negative bacterial strains isolated during the same period in our ICU, showed different patterns. Since four independent events of acquisition of different strains with similar plasmids is unlikely, acquisition of the resistance plasmid by different strains seems the most likely explanation for our observation. The multidrug-resistant (MDR) strains described in our study were intermediately susceptible to polymyxin E. Of note, in our hospital, outside the ICU, polymyxin E is not usually used to treat infections caused by Escherichia coli or Klebsiella pneumoniae. Cefotaxime, the most common systemic antibiotic used in SDD regimens has been reported to be associated with the emergence of ESBL-producing strains and with an increase in ESBL prevalence. In our view, an important part of the problem is the colonization with strains carrying MDR plasmids, as every colonized patient forms a potential source which may facilitate further dissemination of ESBL genes. Although adjustment of the SDD regimen or isolation measures can eventually contain outbreaks, detection of ESBL plasmid-mediated outbreaks is more complex than that of clonal outbreaks. In all studies that evaluated emergence of resistance with SDD wherein cefotaxime is used, there was no optimal screening for ESBLs. Therefore, we suggest that for optimal evaluation of SDD regimens with respect to ESBL-producing strains and emergence of resistance, optimal screening techniques for detection of ESBLs is warranted. Obviously, optimal screening for ESBL production is also needed in ICUs where SDD is not used.