Antibiotic Strategies in the Era of Multidrug Resistance

George Karam; Jean Chastre; Mark H. Wilcox; Jean-Louis Vincent


Crit Care. 2016;20(136) 

In This Article

Links Between Antibiotic Prescribing and Resistance

Most of the time, antibiotics do not create resistance per se. Antibiotic resistance determinants have been circulating within the microbial genome for millennia, largely predating the manufacture and use of antibiotics by human beings.[11,12] This was conclusively demonstrated by D'Costa et al.[12] using targeted metagenomic analyses of rigorously authenticated ancient DNA from 30,000-year-old permafrost sediments and the identification of a highly diverse collection of genes encoding resistance to β-lactam, tetracycline, and glycopeptide antibiotics, very similar to the genes currently expressed by bacteria in healthcare-associated infections. The emergence and dissemination of resistant microorganisms during antimicrobial therapy is thus essentially driven by the selection of the small fraction of naturally resistant bacteria that exist in all microbiota, because of the pressure they exert on the susceptible microorganisms, giving a survival advantage to the cells that are intrinsically resistant to the antimicrobial agent(s) used. In ICU patients, this phenomenon is not only operational at the level of the infected site but also at the level of the digestive tract microbiota and other commensal floras, where the huge number of bacteria present may very rapidly promote the emergence of drug-resistant microorganisms. Even if these bacteria do not themselves cause disease, they can easily share these resistance genes with bacteria that do, through direct exchange of DNA (by conjugation or extrachromosomal plasmid DNA). One recent study confirmed that short exposure to imipenem in ICU patients was followed by a significant increase in carriage of imipenem-resistant Gram-negative bacilli.[13] The risk of acquisition was 5.9 times higher in patients who received only 1–3 days of imipenem treatment compared with controls, and increased to 7.8 times higher in those who received longer treatments.

Numerous studies in human and veterinary medicine have shown a correlation between consumption of antimicrobials and resistance in bacteria isolated from infected humans.[14–19] Recent studies have also confirmed the direct effect of antibiotic use in selecting resistant organisms at the individual level. For example, Malhotra-Kumar et al.[20] demonstrated in a double-blind, randomized trial performed in a large group of healthy volunteers that macrolide exposure for 7 days led directly to the emergence of resistance in the oral streptococcal flora.

In addition to the selection pressure they exert on susceptible bacteria, antibiotics can influence antibiotic resistance through several other mechanisms, including changes in cell permeability and efflux or alterations in the antibiotic target, and horizontal transfer of resistance genes.[21] Many antibiotics, even at very low concentrations that cannot kill susceptible bacteria, induce the formation of reactive oxygen species (ROS), which can damage bacterial DNA, increasing genetic variability.[22] Selection of hypermutable clones is another undesirable consequence that can enhance resistance development.

The potential to select resistant strains of P. aeruginosa with antibiotics that have activity against P. aeruginosa has been demonstrated repeatedly in the medical literature.[23,24] What has evolved from this risk of selecting resistance is the concept of stratifying therapy for infections based on the likelihood of the presence or absence of P. aeruginosa as the etiologic agent of the infection, and this has been a core consideration in multiple clinical guidelines.[25,26]

There are numerous reports of carbapenemase outbreaks linked to various classes of antibiotics. In contrast to establishing a link to only one specific class of antibiotics, there is an evolving body of medical literature suggesting an important relationship between prior antimicrobial therapy and the subsequent identification of carbapenemase-producing bacteria. In a 4-year case–control study of 102 patients, the only covariate independently associated with CRE in all multivariate analyses was the cumulative number of prior antibiotic exposures.[27] A 26-month case–control study (96 ESBL-carbapenem-resistant K. pneumoniae and 55 ESBL-carbapenem-sensitive K. pneumoniae) from Greece identified both prior cumulative exposure to antibiotics and increasing duration of prior treatment as risk factors.[28] Antibiotic treatments shown in this study to be associated with the isolation of carbapenemases were therapy with β-lactam/β-lactamase inhibitor or with a combination of fluoroquinolone and carbapenem. These data are consistent with previous reports that no particular class of antibiotic is the predominant predisposing factor for selection of carbapenemase production, but that it is rather the intensity and duration of antibiotic therapy which are the most important variables in creating the milieu in which carbapenemase-producing bacteria are selected.