This study was conducted from 2006 to 2008 at three LTCFs within the University of Pennsylvania: 1) LTCF #1, a 124-bed facility; 2) LTCF #2, a 240-bed facility; and 3) LTCF #3, a 200-bed facility. As previously described [21], LTCF residents who were colonized initially with FQ-susceptible E. coli (FQSEC) were followed longitudinally for up to 12 months. Serial fecal samples were obtained approximately every 14 days until recovery of FQREC, discharge, or death. The study was approved by the institutional review board of the University of Pennsylvania.

For the present study, we selected for further characterization the initial FQREC E. coli isolate from each of the 37 LTCF residents in the parent study who were documented previously to have developed incident FQREC colonization while under surveillance, i.e., who were colonized initially only with FQSEC and were found subsequently to be colonized with FQREC, +/− FQSEC (21). Additionally, to assess whether the subject’s first detected FQREC isolate arose within the patient from a preceding colonizing FQ-susceptible isolate (i.e., conversion of a clone from FQSEC to FQREC), for each subject the most recent FQ-susceptible E. coli isolate prior to the index FQREC isolate was included for pulsed-field gel electrophoresis (PFGE) analysis if the sampling date was ≤ 30 days prior to the first fecal swab with FQREC.

Given the multi-step nature of the development of FQ resistance (e.g., accumulation of mutations in the quinolone resistance-determining regions [QRDR] of the gyrA and parC genes), isolates with elevated FQ minimum inhibitory concentrations (MICs) may exhibit mutations in FQ target genes [22–25]. Therefore, isolates with elevated FQ MICs are important in evaluating development of FQ resistance. Accordingly, as described previously [21], FQ resistance was defined operationally as a levofloxacin MIC ≥ 0.25 µg/mL. FQ susceptibility was therefore defined as a levofloxacin MIC < 0.25 µg/mL.

Furthermore, LTCF residents have previously been shown to have high rates of gastrointestinal colonization with multiple distinct strains of E. coli [26]. To identify, isolate, and characterize multiple E. coli colonies with diverse FQ susceptibility profiles from a given patient, stool samples were inoculated and replica-plated, as follows: a perirectal swab with freshly (≤ 24 hours) collected stool sample was inoculated by triple-streaking onto a MacConkey agar plate and incubated at 37°C overnight. Single colonies suspected of being E. coli (up to 25 per plate) were replica-plated onto 4 MacConkey agar plates, each supplemented with a different concentration of levofloxacin (8 µg/mL, 1 µg/mL, 0.25µg/mL, no antibiotic). Each colony underwent PYR and indole testing. Indole-positive and PYR-negative colonies were preliminarily considered E. coli and were saved in 30% glycerol stock at −80°C for further analyses.

From each specimen, up to six each presumed FQ-susceptible (levofloxacin MIC <0.25 µg/mL) and FQ-resistant (levofloxacin MIC ≥0.25 µg/mL) E. coli isolates, as available, were analyzed further. Isolates were identified definitively as E. coli and underwent standardized susceptibility testing using the semi-automated Vitek 2 identification and susceptibility system (bioMérieux, Durham, NC). A patient was considered to be colonized with FQREC if at least one stool isolate was confirmed as FQREC.

Evaluation for specific mechanisms of FQ resistance was performed as described previously [27–29]. Overexpression of AcrAB was measured using the organic solvent tolerance (OST) assay [30, 31]. The genetic relatedness of E. coli isolates was determined by PFGE [27], with all profiles analyzed using the Bionumerics v6.6 (Applied Maths, Austin, TX) and interpreted according to established criteria [32]. Strains were defined as those isolates sharing a pulsotype. Pulsotypes were assigned based on > 94% similarity to reference strains within an established PFGE library [33]. Susceptibility testing was performed using the semiautomated Vitek 2 system and interpreted according to Clinical and Laboratory Standards Institute criteria.

Major E. coli phylogenetic group (A, B1, B2, D) was determined by triplex PCR [34]. ST131 status was determined by detection of ST131-specific single-nucleotide polymorphisms (SNPs) in mdh and gyrB [35]. ST131 isolates were screened by PCR for presence of the O25b rfb variant [8] and for membership in the H30 ST131 subclone [6, 36]. The prevalence of the (CTX-M-15-associated) H30Rx subclone [37] was assessed by PCR-based detection of a subclone-specific SNP (G723) within the allantoin protein-encoding gene, ybbW [38].

Fifty extraintestinal virulence-associated genes were detected by multiplex PCR [35, 39]. Virulence scores were calculated as the total number of virulence genes detected, adjusted for multiple detection of the pap (P fimbriae), sfa/foc (S and F1C fimbriae), kpsM II (group 2 capsule), and clb (colibactin) operons. Isolates were regarded as extraintestinal pathogenic E. coli (ExPEC) if positive for ≥ 2 of the following genes: papA and/or papC (P fimbriae); afa/draBC (Dr-family adhesins); sfa/focDE; iutA (aerobactin receptor); and kpsM II (group 2 capsule synthesis) [36].

Baseline demographic and clinical characteristics were extracted from the LTCF medical record using a standardized data abstraction form [21]. Clinical variables included comorbidities, fecal incontinence, bed-bound status, and presence of a urinary or central venous catheter at the time of study enrollment. All antimicrobial therapy received within 30 days of initial detection of FQREC colonization, was documented, as was receipt of immunosuppressive agents.

Among FQREC isolates, associations were explored between E. coli ST131 and 1) mechanisms of FQ resistance, 2) antibiotic susceptibility profiles, and 3) virulence factors. Clinical characteristics were compared between subjects with and without an ST131 FQREC colonizing isolate. Categorical variables were compared using the Fisher exact test, and continuous variables were compared using the Wilcoxon rank-sum test. Odds ratios (OR) and 95% confidence intervals (CI) were calculated to evaluate the strength of associations between ST131 status and clinical variables. All statistical calculations were performed using commercially available software (STATA v13.0; StataCorp LP, College Station, Texas).

E. coli isolates were analyzed from all 37 LTCF residents who were identified as having incident FQREC colonization (levofloxacin MIC ≥0.25 µg/mL) during the study period. Most (n = 24; 65%) subjects had multiple strains of E. coli detected. The median number of distinct E. coli strains identified on testing from each initial FQREC-containing fecal sample for these 24 residents was 2 (interquartile range [IQR], 2, 3), most commonly one FQ-susceptible and one FQ-resistant strain.

The first-identified FQREC isolate was characterized for each of the 37 residents; of these, 29 (78%) were identified as ST131 by dual-SNP PCR (Table 1). The only other ST131 isolate identified among the study subjects was a FQSEC isolate (levofloxacin MIC of 0.047 µg/mL) from a multiply-colonized patient who had one non-ST131 FQ-resistant strain and two FQ-susceptible strains, including this ST131 isolate. Thus, 29 (97%) of the study’s 30 total ST131 isolates were FQREC.

All 29 ST131 FQREC isolates represented the H30 subclone within ST131, and 9 (31%) belonged specifically to the H30Rx subgroup within the H30 subclone. In contrast, the single ST131 FQSEC isolate did not belong to the H30 ST131 subclone.

The 29 ST131 and 8 non-ST131 FQREC isolates did not differ significantly for levofloxacin MIC (median value, 24 µg/mL versus 20 µg/mL, respectively, P = 0.58) (Table 1). ST131 FQREC isolates had a greater number of replacement gyrA mutations compared to non-ST131 FQREC isolates, with a median of 2 (interquartile range [IQR], 2, 2), versus 2 (IQR, 1, 2) (P = 0.01). All isolates with two gyrA mutations exhibited Ser83Leu and Asp87Asn, regardless of ST131 status. Similarly, ST131 isolates had a greater number of replacement parC mutations compared to non-ST131 isolates, with a median of 2 (IQR, 2, 2), versus 1 (IQR, 0, 1) (P < 0.001). The most common combination of replacement parC mutations among ST131 FQREC isolates, present in all but two of the characterized isolates, was Ser80Ile and Glu84Val.

Differences in virulence gene profiles between ST131 and non-ST131 FQREC isolates are shown in Table 2. Virulence genes significantly associated with ST131 included fimH, sat, usp, ompT, and malX. A similar proportion of ST131 and non-ST131 isolates qualified as ExPEC (45% and 50%, respectively; P > 0.99). There was no significant difference in virulence gene scores between ST131 isolates and non-ST131 isolates, with median scores of 9 (IQR, 8, 10) and 8 (IQR, 4, 12), respectively (P = 0.64). The virulence gene profile of the one FQ-susceptible ST131 E. coli isolate included adhesins (iha, fimH, afa/draBC), toxins (hylD, sat, tsh), siderophore receptors (iroN, fyuA, iutA), and other factors (kpsM II, K5, iss, ompT, ibeA, usp, traT, rfc, cvaC), qualifying this isolate molecularly as ExPEC.

Overall, ST131 and non-ST131 FQREC isolates (as defined by a levofloxacin MIC ≥0.25 µg/mL) exhibited no significant differences in antibiotic susceptibility profiles (Table 3). Eight (28%) ST131 isolates were ceftazidime-resistant; four (50%) of these belonged to the H30Rx subgroup within the H30 subclone. The prevalence of ceftazidime resistance was numerically but not statistically significantly higher among H30Rx isolates compared to the other ST131 isolates (44% versus 20%; P = 0.20).

PFGE was performed for 28 of the 29 FQREC ST131 isolates, and pulsotypes were compared to those of previously characterized clinical ST131 isolates from a large, established library [33]. Among the present ST131 colonizing isolates the predominant pulsotypes (defined based on ≥ 94% similarity to an index isolate) were 968 and 808, with 7 and 4 isolates respectively (Figure 1), which also are the leading pulsotypes among the diverse-source clinical ST131 isolates in the reference library [33].

In assessing longitudinal development of resistance, 26 (70%) of 37 subjects had a fecal swab that yielded FQSEC (levofloxacin MIC <0.25 µg/mL) within 30 days prior to the first fecal swab that yielded FQREC (levofloxacin MIC ≥0.25 µg/mL). On PFGE analysis, none of these FQSEC isolates exhibited ≥ 76% profile similarity to the corresponding patient’s subsequent FQREC isolate. Furthermore, none of the FQSEC isolates were ST131.

Among the 37 subjects with incident FQREC colonization, clinical characteristics were compared between the 29 subjects whose new FQREC strain was ST131 and the 8 for whom it was not (Table 4). These two subgroups did not differ significantly for LTCF of residence, presence of various comorbidities, device use, or receipt of various antibiotic agents or classes (P > 0.10 for all comparisons). There was a trend approaching statistical significance toward an association between female gender and having an ST131 incident colonizing FQREC strain (OR, 0.12; 95% CI, 0.01–1.03; P = 0.05).