Different Factors Associated with CTX-M-Producing ST131 and Non-ST131 Escherichia coli Clinical Isolates

Objectives To determine factors associated with CTX-M-producing ST131 Escherichia coli which is the worldwide predominant lineage among CTX-M-producing E. coli isolates. Methods Consecutive inpatients with a clinical sample positive for CTX-M-producing E. coli and considered as cases in a previous 8-month (2008–2009) case-control study performed in ten university hospitals in the Paris area were included in the present sub-population study. Patients with a CTX-M-producing ST131 E. coli clinical isolate were compared with those with a CTX-M-producing non-ST131 E. coli clinical isolate with regard to 66 variables. Variables were first compared using univariate logistic regression, then a multivariate analysis using a backward selection with variables with p-value <0.1 in univariate analysis was carried out. Results Fifty-five patients with a CTX-M-producing ST131 E. coli clinical isolate were compared to 97 patients with a CTX-producing non-ST131 E. coli clinical isolate. Multivariate analysis showed that only previous residence in long term care facilities (OR = 4.4; 95% CI = 1.3–14.7) was positively associated with a CTX-M-producing ST131 E. coli isolate. However, it also showed that regular consumption of poultry products (OR = 0.2; 95% CI = 0.1–0.6), having had at least one device in the preceding 6 months (OR = 0.3; 95% CI = 0.1–0.7) and stay in ICU (OR = 0.2; 95% CI = 0.05–0.8) were negatively associated with isolation of CTX-M-producing ST131 E. coli from clinical samples. Conclusions This study provides more insight into the epidemiological features of ST131 and non-ST131 E. coli producing CTX-M enzymes. It shows, for the first time, that isolation of CTX-M-producing ST131 E. coli from clinical samples is not linked to consumption of various foods and confirms that residence in long term care facilities is a predictor of these isolates.


Introduction
The polyclonal structure of Escherichia coli from clinical and commensal human isolates, and from environmental isolates has clearly been shown by studies recently carried out in the Netherlands (clinical and commensal human isolates and chicken meat isolates), England (clinical isolates) and France (clinical and commensal isolates) on the basis of sequence types (ST) [1][2][3][4][5]. However, some E. coli lineages were identified as predominant in the five above cited studies independent of the source of the isolate or the production of extended-spectrum blactamase (ESBL). In particular, E. coli ST131 was predominant among the clinical and commensal human isolates, producers of ESBL or not. In contrast, it was not identified in ESBLproducing E. coli isolated from chicken meat in the Netherlands. The absence of clone ST131 has also been confirmed recently in Spain among the E. coli isolates contaminating raw chicken meat [6] although another previous Spanish study had found that 7% of retail chicken samples were contaminated by E. coli ST131 [7]. Vincent et al. also had identified E. coli ST131 from retail chicken samples in Canada but at a significant lower prevalence (0.4%) than in Spain [8]. In contrast, isolates of ST10, comprising ESBL and non-ESBL producers, were frequent both among the clinical and commensal human isolates as well as among the meat isolates [1,2,4,5,8]. As shown by the Dutch, Canadian and French studies, the recognized avian pathogenic E. coli ST117 was another predominant lineage among the clinical and meat isolates [1,5,9]. On the other hand, although CTX-M-15 was shown to be the predominant CTX-M enzyme (46%) among the French clinical isolates, it should be stressed that CTX-M-1 was the only ESBL found in the ST117 clinical isolates in France and was the predominant ESBL found among the ST117 meat isolates in the Netherlands [1,5].
These reports suggest that epidemiological differences exist between CTX-M-producing strains of ST131 and non-ST131 clones. Therefore, we sought to analyse characteristics associated with CTX-M-producing E. coli ST131 isolated from clinical samples by performing a sub-population analysis of data collected during a case-control study carried out from November 2008 to June 2009 to determine factors independently associated with a clinical sample positive for a CTX-Mproducing E. coli isolate in ten hospitals of the Paris area [10]. The analysis of the population structure of CTX-M-producing E. coli and non-ESBL-producing E. coli isolates which was performed in addition to the case-control study, was also used as a basis for the present study [1].

Ethics Statement
Written informed consent was obtained from all adult participants and from parents for child participants. The study and the consent procedure were approved by the Ethics Committee of the Groupe Hospitalier Universitaire Nord (Institutional review board NuIRB00006477).

Study Design and Participants
All consecutive inpatients with a clinical sample positive for CTX-M-producing E. coli and considered as cases in a previous 8month (2008)(2009)) case-control study performed in ten university hospitals in the Paris area were included in the present subpopulation study [10]. Patients with a clinical sample positive for CTX-M-producing ST131 E. coli (n = 55) were compared with those with a clinical sample positive for CTX-M-producing non-ST131 E. coli (n = 97) with regard to 66 characteristics collected during the case-control study, including basic demographic data, patient's lifestyle (housing, travel abroad, diet, pet, sport practice…), medical history (hospitalisation and invasive devices in the preceding six months, antibiotic in the preceding month, co-morbidity…) and data on the current hospitalisation (hospitalisation wards, invasive devices, antibiotic regimens …). The 97 non-

Statistical Analysis
Variables were first compared using univariate logistic regression and odds ratio (OR) and 95% confidence interval (CI) were estimated. We next used a multivariate analysis using a backward selection with variables with p-value ,0.1 in univariate analysis. Pvalues were assessed at the 0.05 level. All statistical analyses were performed with SAS software, version 9.3 (SAS Institute, Cary, North Carolina).

Discussion
E. coli ST131 has been shown to be a worldwide predominant clone among extra-intestinal pathogenic isolates but also among the human commensal flora [2,4,[10][11][12]. Interestingly, it was found to be almost the only lineage among clinical isolates of group B2 E. coli that produced CTX-M enzymes [1]. It displayed a higher ability to colonize the digestive tract and a lower level of virulence in various animal models in comparison with reference group B2 urinary pathogenic E. coli strains (CFT053, J536 and HT7) [13][14][15]. Therefore, better knowing the epidemiology of clone ST131, which appears to be a very peculiar group B2 lineage, especially among isolates producing CTX-M enzymes, is of interest due to its worldwide success. The present prospective study investigated which factors among 66 studied were associated with those of ST131 E. coli clinical isolates that produce CTX-M enzymes. Among the various types of food products analysed, it was found, for the first time, that consumption of poultry meat at least twice a week is a factor inversely associated with isolation of a CTX-M-producing ST131 E. coli clinical isolate among the CTX-M-producing E. coli clinical isolates. In other words, it means that consumption of poultry meat was associated with isolation of CTX-M-producing E. coli that did not belong to ST131. This finding is of importance with regard to the debate on the potential food-borne source of E. coli ST131, notably those producing CTX-M enzymes [8,16]. Poultry meat was suggested as a source of E. coli ST131 on the basis of two studies published in 2010 because E. coli ST131 has been isolated from poultry meat samples [7,8]. The most recent studies conducted in the Netherlands and in Spain challenged this hypothesis as they failed to isolate CTX-Mproducing E coli ST131 from chicken meat samples [3,6]. Overall, the results of our study are in accordance with the fact that E. coli ST131 has not been identified among ESBL-producing E. coli isolated from retail chicken meat on the contrary to other lineage [3,5,6,17]. Although, there are very few studies on the population structure of ESBL-producing E. coli isolates from poultry meat, it is noteworthy that, among the CTX-M-producing non ST131 E. coli clinical isolates, some dominant clonal groups (ST10, ST117 and ST354) are commonly identified from chicken meat [3,8,18]. Interestingly, ST167 and ST648, the two highest dominant clonal groups after ST10 among the CTX-M-producing non-ST131 E. coli clinical isolates had been identified among ESBL-producing E. coli isolates from Spanish poultry farms and from birds of prey from Germany and Mongolia [19,20]. In summary, the dominant non-ST131 clonal groups in our population are clonal groups commonly identified in avian populations. The only factor positively associated with isolation of CTX-Mproducing E. coli ST131 from clinical samples was residence in LTCF before inclusion in the study. Rooney et al. showed that a high proportion of people living in such settings in England had digestive tract colonization with ESBL-producing E. coli ST131 [21]. Of note, the first identification of CTX-M-15-producing E. coli ST131 in France was achieved from patients in LTCFs [22,23]. More recently, Banerjee et al. conducted a retrospective study in all healthcare settings in Olmsted County (Minnesota) and found that LTCF residence was a factor independently associated with E. coli ST131 [24]. Overall, three studies conducted in three different developed countries have found a link between LTCF residence and E. coli ST131. This might suggest that human crosstransmission is a key factor in the dissemination of CTX-Mproducing E. coli ST131.
Although the proportion of hospital-acquired (isolation after 48 h of hospitalisation) CTX-M-producing ST131 and non-ST131 E. coli isolates was high and not significantly different (63.6% vs 56.2%: P = 0.2) and the patients infected by either E. coli ST131 or E. coli non-ST131 did not differ with regard to Mac Cabe score, we found that presence of invasive devices in the preceding six months and stay in ICU before study inclusion were inversely associated with isolation of CTX-M-producing E. coli ST131. It suggests that isolation of CTX-M-producing non-ST131 E. coli from clinical samples is more likely to be healthcare-related. Such results seem to be in contradiction with those obtained by Banerjee et al. [24]. Indeed, they found that E. coli ST131 is linked to healthcare and hospital acquisition. However, we noted that this link was identified by Banerjee et al in the univariate and not in the multivariate analysis that they carried out. Finally, we were not able to link travel abroad, notably in Africa and India, to isolation of CTX-M-producing E. coli ST131 from clinical samples probably because of the lack of power regarding this association in our study [25].
In conclusion, this study provides more insight into the epidemiological features of ST131 and non-ST131 E. coli producing CTX-M enzymes. It shows, for the first time, that isolation of CTX-M-producing E. coli ST131 from clinical samples was not linked to consumption of specific foods and confirms that residence in long term care facilities is linked to these isolates. Further studies are required to know whether our results are also relevant for E. coli ST131 not producing CTX-M enzymes.