Editorial Commentary: Extended-Spectrum -Lactamase-Producing Escherichia coli in the United States: Time to Rethink Empirical Treatment for Suspected E. coli Infections?

(See the Major Article by Doi et al on pages 641–8.) In this issue of Clinical Infectious Diseases, Doi et al report the results of a prospective observational study conducted among 5 hospitals and their affiliated clinics throughout the United States aimed at estimating the occurrence of community-acquired infections due to extended-spectrum β-lactamase (ESBL)–producing Escherichia coli [1]. Similar to reports from Europe [2, 3], the vast majority of infections caused by ESBL-producing E. coli were referable to the urinary tract, from cystitis, pyelonephritis, or bacteremia with a urinary source. This is what we might expect: E. coli is the most common cause of urinary tract infection (UTI) in the hospital and the community, and UTI is the most common bacterial infection [4]. Doi et al only analyzed ESBL-producing E. coli; therefore, it is important to put their conclusions into context [1]. They screened 13 279 unique E. coli isolates in order to find 523 that were ESBL-producing. This is 3.9% of the total isolates, of which about half (292 [55%]) were community acquired of these, 64.2% were healthcare associated. For comparison, a Canadian surveillance study testing E. coli that caused incident bloodstream infections between 2000 and 2010 identified 4% as ESBL producers; 78.6% were community acquired, and 67.7% of these were healthcare associated [5]. Unfortunately, Doi et al do not report what proportion of the 13 279 were community acquired rather than healthcare associated [1]. Of the 292 ESBL producers, the majority (54.2%) belonged to multilocus sequence type ST131 or a related sequence type carrying CTX-M–type ESBL genes. This led them to assert that community-associated ESBL-producing E. coli infections in the United States are driven in large part by clonal expansion of ST131-carrying CTX-M–type ESBL genes. Although this assertion would have been more convincing if all of the ESBL-producing isolates had been sequenced and analyzed using phylogenetics, and the prevalence of ST131 among E. coli not producing ESBL had been estimated, it is clear that half of the ESBL-producing E. coli isolates in this study shared a common lineage. However, the other half did not. Because ESBL production can be acquired by horizontal gene exchange, constraining ST131 will help, but not stop, the spread of ESBL-producing E. coli. The virulence and transmissibility of ESBL-producing E. coli is unknown, but we do know a little about ST131. Gibreel et al compared the pathogenic potential of ST131 to other uropathogenic E. coli; ST131 had significantly fewer known virulence factors [6]. Similarly, Lavigne et al [7] showed that ST131 was less virulent than other uropathogenic E. coli strains in a zebrafish embryo model. ST131 does, however, seem to have higher metabolic potential [8]. It is possible that the virulence and/or transmissibility of ESBL-producing E. coli is the same or even lower than other pathogenic E. coli, but because of the antibiotic resistance profile, the initial treatment is more likely to be inadequate, leading to more severe disease. Doi et al found several factors associated with healthcare-associated episodes of community-acquired infection caused by ESBL-producing E. coli: cardiovascular diseases, chronic renal failure, dementia, solid organ malignancy, and hospitalization within the previous 12 months [1]. Unfortunately, they did not stratify by site of infection, which might have given some insight into whether these factors are markers of increased susceptibility to infection once colonized, or increased exposure to ESBL-producing E. coli. A similar study conducted in Switzerland using 123 ESBL-producing E. coli isolated from the urine found no difference in patient comorbidities, with the exception of nonurological malignancies, which were more common among patients with healthcare-associated UTIs. The proportion of patients with complicated UTI was the same among community-acquired and healthcare-associated UTIs due to ESBL-producing E. coli [9]. To constrain spread of ESBL-producing E. coli will require reducing factors positively selecting for ESBL. Because ESBL-producing E. coli strains are often resistant to multiple drugs, particularly fluoroquinolones [3, 9, 10], the selection pressure may not be for ESBL per se but for one of the other resistance mechanisms found with it. Individuals with community-acquired UTI due to ESBL-producing E. coli are 4.1 times more likely to have previously used fluoroquinolones. While the associations with using third- and second-generation cephalosporins are much higher, as might be expected because ESBL has a direct effect on cephalosporins, ESBL has no effect on fluoroquinolones [11]. In the study by Doi et al, the vast majority (87.5%) of isolates were resistant to both levofloxacin and ciprofloxacin [1]. The antibiotic resistance profile and urinary origin of the majority of isolates suggests that prescribing practices for community-acquired UTIs are likely an important positive selection factor for ESBL-producing E. coli. While fluoroquinolones are effective for UTI treatment [12], the increasing resistance to fluoroquinolones among uropathogenic E. coli [4], combined with the frequent co-occurrence of fluoroquinolone resistance with ESBL production [1, 3], strongly suggests that the fluoroquinolones should be preserved for infections that have the potential to cause serious morbidity and mortality, that is, pyelonephritis rather than cystitis. A joint guideline from the Infectious Diseases Society of America (IDSA) and European Society for Microbiology and Infectious Diseases recommends treating uncomplicated UTIs with nitrofurantoin, trimethoprim-sulfamethoxazole, fosfomycin, or pivmecilliman [13]. Fluoroquinolones or β-lactams are only recommended if the others are not available or the patient has a history of allergy or tolerance to the first-line therapies. Following urine culture and susceptibility testing, ciprofloxacin is recommended for treating pyelonephritis. The incidence of lower UTI is about 3 orders of magnitude higher than that of pyelonephritis [4, 14]; therefore, the selection pressure applied from treating pyelonephritis is considerably less than that from treating lower UTI. Selection pressure might also be reduced by not treating most asymptomatic bacteriuria. In a related guideline, also from IDSA, treatment of asymptomatic bacteriuria is not recommended for premenopausal nonpregnant women, diabetic women, the elderly living either in the community or in an institution, persons with spinal cord injuries, or catheterized patients while the catheter is in place [15]. Unfortunately, therapy for human diseases is not the only factor selecting for ESBL-producing E. coli. ESBL-producing E. coli has been found in healthy pigs, cattle, and poultry in China [16] and in wildlife, possibly due to manure spread of ESBL-producing E. coli [17]. The same ESBL genes found among E. coli from poultry and retail chicken meat in the Netherlands were found among human clinical isolates collected during the same time period [18]. Therefore, antibiotic use in feed animals may be further driving the spread of ESBL genes. It is reassuring that among the ESBL-producing E. coli strains identified by Doi et al, all—regardless of source of infection—were sensitive to nitrofurantoin [1]. If practitioners follow the IDSA and European Society for Microbiology and Infectious Diseases treatment guidelines [13], at least for now, it will not be necessary to rethink empirical treatment for community-acquired infections where E. coli is suspected. However, antibiotic resistance is a moving target. Ongoing local surveillance for antibiotic resistance combined with antibiotic stewardship is essential to minimize inadequate therapy and avoid adverse health outcomes. Limiting antibiotic use in feed animals and empirical antibiotic therapy to treat mild, and generally self-limiting, human infections might also be prudent. Potential conflicts of interest. Author certifies no potential conflicts of interest. The author has submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.