Escherichia coli Sequence Type 131 H30 Is the Main Driver of Emerging Extended-Spectrum-β-Lactamase-Producing E. coli at a Tertiary Care Center

The ever-rising prevalence of resistance to first-line antibiotics among clinical Escherichia coli isolates leads to worse clinical outcomes and higher health care costs, thereby creating a need to discover its basis so that effective interventions can be developed. We found that the H30 subset within E. coli sequence type 131 (ST131-H30) is currently, and has been since at least 2004, the main E. coli lineage contributing to key resistance phenotypes—including extended-spectrum-beta-lactamase (ESBL) production, fluoroquinolone resistance, multidrug resistance, and dual ESBL production-plus-fluoroquinolone resistance—at a United States tertiary care center with a rising prevalence of ESBL-producing E. coli isolates. This identifies ST131-H30 as a target for diagnostic tests and preventive measures designed to curb the emergence of multidrug-resistant E. coli isolates and/or to blunt its clinical impact.

In 2014, Harbor-UCLA Medical Center (Los Angeles, CA) noted a steadily rising prevalence of ESBL production among E. coli clinical isolates in its antibiograms over the six preceding years, from 8.7% (297/3,405) (in 2008) to 20.8% (829/3,989) (in 2013), with most ESBL-producing strains being multidrug resistant (MDR) and, specifically, having a 90% prevalence of fluoroquinolone resistance (L. Miller, unpublished data). Of note, these antibiograms did not include duplicate isolates of the same organism with the same antimicrobial susceptibility profile from the same patient within 2 weeks. Here, we sought insights into this ESBL emergence; i.e., whether it was most likely due to extensive horizontal transfer of ESBL-encoding determinants or the expansion of a single or multiple ESBL-producing lineages and, if so, which. We also sought to understand the clonal background of resistance generally.
Clonal distribution of source and resistance. The major phylogenetic groups accounted minimally for variation in source and antimicrobial resistance (Table 1). In contrast, nearly all clonal groups or subclones were significantly or borderline significantly associated with at least one source/resistance variable (Table 3). ST131-H30, with its H30Rx subclone, exhibited among the highest values for the prevalence of resistance and accounted for the greatest share of resistant isolates, including 47% of ESBL producers and of those with bla CTX-M group 1/bla CTX-M-15 or resistance to ESCs, fluoroquinolones, gentamicin, piperacillin-tazobactam, or tobramycin. In contrast, the non-H30 ST131 isolates generally exhibited low values for the prevalence of resistance that, with few exceptions, did not differ from those of the rest of the population. Within ST131-H30, the prevalence of ESBL production and bla CTX-M-15 was greatest for H30Rx. Similarly, STc141, despite its small number of isolates (n ϭ 3), exhibited multiple significant or borderline-significant associations with source and resistance, and STc405 was borderline significantly associated with trimethoprim-sulfamethoxazole (TMP-SMX) resistance. In contrast, for one or more agents, STc95, STc73, and STc69 were associated negatively with resistance.
Resistance scores and MDR status were distributed accordingly (Table 4). Whereas no major phylogenetic group was associated significantly with either variable, ST131-H30 and its H30Rx subclone exhibited significantly higher resistance scores and MDR prevalence values than did other isolates. Notably, ST131-H30, with its 100% prevalence of fluoroquinolone resistance, accounted for most isolates (53%) that were dually ESBL producing and fluoroquinolone resistant. In contrast, the diverse other clonal groups   that accounted for the remaining 47% of dually resistant isolates each contributed only a single isolate (not shown). Among these other clonal groups was STc14 (n ϭ 11), which, despite resembling ST131-H30 fairly closely for the prevalence of MDR status (55%) and fluoroquinolone resistance (91%), had only one ESBL-producing representative. As for other notable features of different clonal groups, STc95 and STc73 isolates   each exhibited significantly lower resistance scores and a numerically lower MDR prevalence than did other isolates.
Historical ESBL-producing isolates. The temporal stability of clonal relationships was assessed among the institution's archived ESBL-producing bloodstream isolates (2004 to 2011) ( Table 5). Over this 7-year period, the prevalence of extended-spectrumcephalosporin resistance among the institution's E. coli clinical isolates rose steadily, from 3% (among 3,106 total isolates in 2004) to 13% (among 3,463 total isolates in 2011), according to the institution's cumulative antibiogram. Clonal analysis of prospectively archived ESBL-producing bloodstream isolates from these years showed that, notwithstanding minor fluctuations, the overall clonal distribution remained fairly stable, with ST131-H30 and H30Rx predominating (46% and 37%, respectively). No other single clonal group, including non-H30 ST131 isolates, was nearly so prominent among the historical ESBL isolates as ST131-H30.

DISCUSSION
The findings of this study, which analyzed 109 recent and 41 historical clinical E. coli isolates from a United States tertiary care center with a rising prevalence of ESBLproducing E. coli isolates, confirm the strong linkage of specific E. coli clonal groups with antimicrobial resistance (4,5,9). Here, lineage-specific associations were much stronger and more numerous than associations involving major E. coli phylogenetic groups, which illustrates the value of "drilling down" to biologically relevant subspecific taxa, in contrast to the standard clinical laboratory practice of identification only to the species level (5,9). Importantly, the observed lineage-specific associations included both exceptionally high and low resistance prevalences and scores (5,9). Such lineagespecific susceptibility data, if combined with rapid determination of clonal group (9), could be used clinically for selecting optimal patient-specific antimicrobial therapy, conceivably even at the point of care.
The high prevalence of ST131-H30 and its H30Rx subclone, both overall and especially among resistant isolates, has potentially important practical implications. First, the prominence of H30Rx among the present and historic ESBL-producing isolates (35% to 37%) suggests that the institution's recent surge in ESBL-producing E. coli, while polyclonal, largely represents the further expansion of this pandemic MDR lineage, rather than new emergence of other ESBL-producing lineages or widespread horizontal gene transfer. This implies that focused attention to this single strain,  including through rapid detection (9), transmission blockade, and/or reservoir elimination, could help in reducing ESBL infections. Notably, our archival findings from 2004 to 2011 contrast with similar data from Calgary from 2000 to 2010 that show a progressively increasing ST131 fraction among ESBL-producing E. coli bloodstream isolates (10); this suggests that the dynamics of clonal emergence may vary geographically. Second, apart from ESBL production, ST131-H30 accounted for approximately half of the study's 2015 fluoroquinolone-resistant isolates, similar to results from other centers (5,8). This suggests that efforts to reduce fluoroquinolone resistance-related morbidity and costs should include attention to this strain. The clinical importance of the combined ESBL production and fluoroquinolone resistance phenotype, which here was strongly associated with ST131-H30 and H30Rx, has been noted recently for patients with acute pyelonephritis (3).
Additionally, STc14 (O75 associated), also from group B2, accounted for two-thirds of the present non-ST131-H30 fluoroquinolone-resistant isolates. STc14 has been associated previously with antimicrobial resistance, including to ampicillin (11) and fluoroquinolones (12). Interestingly, here STc14 did not contribute appreciably to the clonal expansion of ESBL-producing strains, indicating that while ESBL production is associated closely with fluoroquinolone resistance, the opposite is not necessarily true. The discovery of how and why ST131-H30 and STc14 have diverged so markedly with respect to their resistance profiles, both from other group B2 lineages and from each other, could prove valuable.
Also notable were three group D-associated lineages-STc69, STc405, and STc31. STc69, initially called "clonal group A," appeared in the 1990s as a widespread cause of TMP-SMX-resistant UTIs (13,14). Here, STc69, although prominent (as the next-mostprevalent clonal group after ST131 and slightly more prevalent than ST131-H30), was associated with antimicrobial susceptibility. STc405, noted recently as an emerging MDR clone (15), appeared comparatively benign, except for a borderline high prevalence of TMP-SMX resistance. In contrast, STc31 (also known as the O15:K52:H1 clonal group), the first MDR E. coli lineage known to cause a community-wide outbreak (16,17), was absent. This aligns with STc31's usually minor contribution to recent E. coli study populations, albeit sometimes in association with fluoroquinolone resistance (4,18).
The study has limitations. First, ESBL-encoding genes were detected by PCR but not sequenced, which leaves uncertainty as to the specific bla variant(s) present in several isolates; however, such data would be unlikely to change the study's main conclusions. Second, epidemiological data were unavailable, which precluded analyses of host characteristics, clinical manifestations, and outcomes in relation to clonal background (19). In that regard, it is possible that the emergence of ESBL-producing E. coli at this institution was due more to expansion of the at-risk host population (e.g., elderly, compromised, and antibiotic-exposed patients) than to special bacterial strains or traits (19). Third, the historical isolates were all blood isolates, possibly biasing the comparison of historical versus recent isolates; however, most previous studies have not found clonal distribution to vary significantly in relation to specimen type (e.g., see reference 8). Fourth, the use of multiple comparisons risked finding spurious associations by chance alone, whereas the small group size for some comparisons risked missing true associations due to low statistical power.
The study also has strengths. These include the recent origin of the 2015 isolates, the inclusion of a historical comparison group obtained from 2004 through 2011, the detailed clonal typing, and the comparisons of clonal background with clinically relevant resistance phenotypes and genotypes.
In summary, among 109 recent and 41 archival clinical E. coli isolates from a public tertiary care center in Los Angeles, we identified strong clonal associations with antimicrobial resistance, especially for ST131-H30 and its ESBL-associated H30Rx subclone. The predominance of H30Rx within a polyclonal ESBL expansion likely underlies the recent surge in ESBL-producing E. coli isolates at this center. These clonal associ-ations support the use of rapid diagnostics for improved prescribing and focused infection prevention efforts.

MATERIALS AND METHODS
Isolates and susceptibility testing. Harbor-UCLA Medical Center is a 400-bed tertiary care public hospital; its laboratory processes specimens from patients in the hospital and from the emergency department and dozens of affiliated general and specialty clinics in the center and surrounding community. For this study, the Clinical Microbiology Laboratory provided 109 consecutive single-patient E. coli isolates from 2015, accompanied by data regarding antimicrobial susceptibility (obtained using Clinical Laboratory Standards Institute interpretive criteria and the Vitek II instrument [bioMérieux, Durham, NC]), specimen type, and inpatient/outpatient source. Additionally, the Infection Control Department provided all 41 available archived ESBL-producing E. coli bloodstream isolates from 2004 to 2011, accompanied by the corresponding annual antibiograms. These archival isolates were drawn from the Infection Control Department's prospectively assembled collection of all sterile site isolates from the Clinical Microbiology Laboratory, which were saved regardless of species and susceptibility profile.
ESBL production was inferred from the enhancement of ESC susceptibility by clavulanate (20). Intermediate results were considered resistant. The resistance score was the number of drug classes to which resistance was detected, counting penicillins and cephalosporins separately. Isolates with resistance scores of Ն3 were considered multidrug resistant (MDR) (21).