Rapid and Extensive Expansion in the United States of a New Multidrug-resistant Escherichia coli Clonal Group, Sequence Type 1193

Abstract

We describe the rapid and ongoing emergence across multiple US cities of a new multidrug-resistant Escherichia coli clone—sequence type (ST) 1193—resistant to fluoroquinolones (100%), trimethoprim-sulfamethoxazole (55%), and tetracycline (53%). ST1193 is associated with younger adults (age <40 years) and currently comprises a quarter of fluoroquinolone-resistant clinical E. coli urine isolates.

Antibiotic resistance is rising globally at an alarming rate. Urinary tract infections, caused primarily by Escherichia coli, are a major reason for antibiotic use, and fluoroquinolones (FQs) are among the most commonly used agents. One decade ago, an FQ-resistant (FQ-R) multilocus sequence type (ST) of E. coli, ST131, was found to have emerged globally in a pandemic fashion. Subsequent studies showed that the ST131 pandemic is driven by the H30 subclone, which emerged in the late 1990s, likely on the US East Coast [1]. No other E. coli clonal group has yet been identified that matches the rapid expansion and dominance of H30, which limits our ability to pinpoint the mechanisms whereby successful drug-resistant E. coli strains can emerge and disseminate. However, starting in 2012, reports from individual hospitals in the United States, Australia, China, South Korea, and Norway have documented the frequent occurrence among FQ-R E. coli isolates of the clonal group ST1193 [2–6]. Here, we analyze whether, besides H30, any other clonal expansion is evident among FQ-R E. coli isolates collected in the course of a multicenter surveillance study.

METHODS

As part of a multicenter surveillance study, 6349 consecutive clinical E. coli isolates from urine (96.0%), blood (3.5%), or wounds (0.5%) were obtained in 2016–2017. The study involved 9 clinical microbiology laboratories (sites) in 4 geographically dispersed cities, including Seattle, Washington (Kaiser Permanente Washington [KPWA], Seattle Children’s Hospital, and Harborview Medical Center); Los Angeles, California (Keck and Los Angeles County + University of Southern California medical centers); Minneapolis, Minnesota (Veterans Affairs and Hennepin medical centers); and New York, New York (New York University Langone Medical Center and Langone Brooklyn Hospital). Susceptibility to 12 antibiotics, representing 8 drug classes, and production of extended-spectrum β-lactamases (ESBLs) were tested by disk diffusion according to Clinical and Laboratory Standards Institute guidelines [7], with intermediate isolates designated as resistant. FQ-R isolates were analyzed further for clonal identity by using a single-nucleotide polymorphism (SNP)–based clonotyping test (7-SNP test) [8] and fumC/fimH sequence typing, which identified the isolate’s ST-fimH clonal group [9]. Mutations in the quinolone resistance–determining regions (QRDRs) of gyrA and parC were identified by sequencing. Presence of the fimH64 allele was identified in quantitative polymerase chain reaction (PCR) using a designed fimH64 SNP-specific primer, H64-F2, 5ʹ-GAACGGATAAGCCGTGACT-3ʹ (forward), in combination with previously described primer 488/483-R, 5ʹ-TCT GCGGTTGTGCCGGATAGG-3ʹ (reverse) [8], which yields a 316-bp PCR product. EnteroBase was used to identify the differences between the ST14 and ST1193 multilocus sequence type (MLST) alleles (https://enterobase.warwick.ac.uk/species/ecoli).

At each study site, the prevalence of H30 and ST1193 was compared statistically using the likelihood ratio test. To compare current and historic prevalence values for H30 and ST1193, we used urine E. coli isolates from a previously described study done in 2011 at 4 of the present study sites (KPWA, Harborview, Seattle Children’s Hospital, and Veterans Affairs) according to the same protocol as used here [10]. Temporal changes in clonal group prevalence among FQ-R E. coli isolates were estimated using multiple logistic regression, with by-site adjustment for either 4 sites (2011 to 2016–2017 analysis) or 8 sites (2016–2017 analysis). Association of FQ-R E. coli isolates with age was estimated using data from the above-described 2011 study and a 2015 study performed at KPWA [10, 11]. A 2-sample Kolmogorov-Smirnov (K-S) test was used to evaluate differences in age distribution for patients with H30 vs ST1193; the binary age cutoff was chosen based on the age with the largest K-S statistic [12]. Associations of clonal background with age group were assessed using a χ² test; multiple logistic regression was used to account for possible confounding due to different studies. Resistance prevalence among H30 and ST1193 isolates from 9 sites (2016–2017) was compared using a χ² or 2-tailed Fisher exact test, as appropriate. Statistical analysis was performed using Stata/IC 14.0 software (StataCorp, College Station, Texas).

RESULTS

Of the 6349 total E. coli study isolates, 1314 (20.7%) were FQ-R. The FQ-R isolates represented 45 clonal groups overall (10–28 per site). At each site, the most prevalent clonal group was H30 (per-site mean, 45.4%), and the second most prevalent was ST1193 (per-site mean, 23.2%) (Table 1). Although H30 was significantly more prevalent than ST1193 at 7 of 9 sites (P < .05)—including at least 1 per city—at the remaining 2 sites, each in a different city, the prevalence difference between H30 and ST1193 was nonsignificant (Table 1). Collectively, H30 and ST1193 accounted on average for 68.6% of FQ-R isolates per site (range, 59.1%–79.8%). By contrast, the next 3 most prevalent 7-SNP clonotypes, corresponding to ST405-fimH27, ST10-fimH54, and ST131-fimH41, each accounted for only approximately 3% of FQ-R isolates and were not all found at some of the study sites (not shown).

Table 1.

Comparison of Fluoroquinolone-resistant (FQ-R) Escherichia coli Clones H30 and Sequence Type 1193 by Prevalence Among Other FQ-R E. coli by Collection Site, Year, Patients’ Age, Clinical Source, and Prevalence of Resistance to Other Antibiotics

Category	Specific Group (Total No. of Isolates)	H30a	ST1193a	P Valueb
Siteb	Total (N = 1314)c	589 (45.5 ± 5.8)	301 (23.2 ± 7.2)	<.001
	KPWA (n = 308)	121 (39.3)	72 (23.4)	<.001
	Harborview Medical Center (n = 183)	86 (47.0)	60 (32.8)	.031
	Seattle Children’s Hospital (n = 163)	69 (42.3)	37 (22.7)	.002
	Minneapolis VA Medical Center (n = 147)	74 (50.3)	20 (13.6)	<.001
	Hennepin Medical Center (n = 152)	79 (52.0)	38 (25.0)	<.001
	Keck USC Medical Center (n = 139)	63 (45.3)	21 (15.1)	<.001
	LAC + USC Medical Center (n = 93)	33 (35.5)	22 (23.7)	.14
	NYU Langone Medical Center (n = 80)	42 (52.5)	14 (17.5)	<.001
	NYU Langone Brooklyn Hospital (n = 49)	22 (44.9)	17 (34.7)	.42
Yearb,c,d	2011 (n = 209)	107 (52.8 ± 5.3)	9 (3.4 ± 1.3)	<.001
	2016–2017 (n = 801)	350 (44.7 ± 2.4)	189 (23.1 ± 3.9)
	2016 (n = 676)	303 (45.8 ± 2.8)	124 (18.4 ± 1.6)	.018
	2017 (n = 589)	264 (46.1 ± 2.7)	160 (25.9 ± 3.2)
Age, yb	18–40 (n = 70)	28 (40.0)	9 (12.9)	.002
	>40 (n = 185)	116 (62.7)	8 (4.3)
Specimenb	Urine (n = 1259)	552 (43.8)	293 (23.3)	.023
	Blood (n = 47)	31 (66.0)	6 (12.8)
Resistance to antibioticsa,b	Amoxicillin-clavulanate	296 (50.3)	75 (24.9)	<.001
	Cefazolin	256 (43.6)	38 (12.7)	<.001
	TMP-SMX	276 (46.9)	163 (54.2)	.019
	Nitrofurantoin	39 (6.6)	19 (6.3)	.96
	Imipenem	1 (0.2)	1 (0.3)	.78
	Tetracycline	262 (44.7)	157 (52.2)	.016
	ESBL producer	190 (32.4)	24 (8.0)	<.001

Data are presented as No. (%) unless otherwise indicated.

Abbreviations: ESBL, extended-spectrum β-lactamase; KPWA, Kaiser Permanente Washington; LAC, Los Angeles County; NYU, New York University; ST, sequence type; TMP-SMX, trimethoprim-sulfamethoxazole; USC, University of Southern California; VA, Veterans Affairs.

^aFor each category except resistance to antibiotics, the percentage of isolates belonging to a specific clone is given in parenthesis; for resistance to antibiotics other than FQ the number of resistant isolates within a specific clone is given with % from all isolates belonging to this clone in parenthesis.

^b P values indicate significance of the difference between H30 and ST1193 characteristics, estimated in likelihood ratio test for prevalence by site; in logistic regression with adjustment by site for time-dependent changes; in χ² test for association with specific age group or specific specimen; in χ² or, if appropriate, 2-tailed Fisher exact test for prevalence of resistance to other antibiotic classes.

^cFor the total prevalence, the mean ± standard error of the by-site prevalence (%) is given in parentheses.

^dPrevalence by year is calculated for 4 sites when 2011 is compared to combined 2016–2017, and for 8 sites when 2016 is compared to 2017.

Between 2016 and 2017, the prevalence of H30 did not change significantly, either overall (means, 45.8% [2016] vs 46.1% [2017]) (Table 1) or at any single site (Supplementary Table 1). By contrast, the prevalence of ST1193 increased both overall (from 18.4% [2016] to 25.9% [2017]; P < .001) and at 6 of 8 sites.

At the 4 sites that provided both historical (2011) and recent (2016–2017) FQ-R isolates (see Methods), between 2011 and 2016–2017 the prevalence of H30 did not change significantly, either overall (average by-site prevalence 52.8% in 2011 vs 44.7% in 2016–2017; P = .06) (Table 1) or at individual sites (Supplementary Table 1). By contrast, during this interval ST1193 exhibited a 7-fold overall prevalence increase, from negligible in 2011 (3.4%) to 23.4% in 2016–2017 (P < .001), and a significant increase at each of the 4 sites (P < .05).

According to the retrospective analysis of 2 previous studies (see Methods), patients with ST1193 were significantly younger on average than patients with H30 (P = .034). The largest K-S statistic (D = .339) was around age 40, with ST1193 isolated significantly more frequently from patients aged 18–40 years than H30 (P = .002; Table 1) or other FQ-R isolates combined (61 of 238 patients [34.5%]; P = .015). With multivariable adjustment by the study, ST1193 remained highly associated with younger age in comparison to H30 (odds ratio, 4.3; P = .007).

Among the present 1306 FQ-R isolates, we compared the frequency of H30 and ST1193 in relation to urine vs blood source. Whereas H30 was 1.5 times more prevalent among blood isolates than urine isolates, ST1193 exhibited an opposite association, being twice as prevalent among urine isolates as blood isolates (Table 1). (Due to the 2016–2017 study design, clinical source could not be adjusted for age or other patient’s conditions.)

As compared with H30 isolates, ST1193 isolates were less frequently resistant to amoxicillin-clavulanate and cefazolin or produced ESBLs, but more frequently were resistant to trimethoprim-sulfamethoxazole and tetracycline (Table 1). Despite these differences, both clonal groups exhibited a similarly high prevalence of multidrug resistance, a similarly low prevalence of nitrofurantoin resistance, and nearly universal susceptibility to imipenem.

Sequence analysis of gyrA and parC among 102 randomly selected ST1193 isolates revealed QRDR mutations in both genes (S83L and D87N in gyrA, S80I in parC), indicating that, as with H30, FQ resistance within ST1193 is of chromosomal origin. Furthermore, according to the MLST database, ST1193 belongs to the ST14 clonal complex within E. coli phylogroup B2 and, across the 7 MLST loci, differs from ST14 by only 1 SNP (a nonsynonymous g551a mutation in icd, producing a G184D amino acid substitution), indicating that ST1193 is a recent evolutionary derivative of ST14. Members of ST14, in contrast to ST1193, are almost all fluoroquinolone susceptible (data not shown). Furthermore, according to fumC/fimH sequence typing, ST1193 carries fimH allele number 64, which differs from allele number 27 of ST14 by only 1 SNP (c311t, producing a P104L substitution). The designed fimH64-specific PCR test (see Methods) was positive for all 75 randomly selected ST1193 isolates, but negative for 623 randomly selected FQ-susceptible isolates from the current study. Thus, fimH64 could be a highly specific marker for ST1193, a clonal group that might be entirely FQ-R.

DISCUSSION

This multicenter molecular surveillance study of FQ-R E. coli clinical isolates establishes the widespread occurrence of ST1193 across the United States as an important and growing contributor to the FQ-R E. coli population. As there are already reports of ST1193 being isolated in hospitals in Europe and Asia, it is likely to be a pandemic clonal group similar to H30.

The emergence of ST1193 represents the only known large clonal expansion of FQ-R E. coli isolates apart from the global spread of ST131 H30 over the last 2 decades. The rise of ST1193 apparently began more recently than that of H30, and in contrast to the H30 expansion, which appears to have plateaued, could still be ongoing rapidly.

Clinically, ST1193 differs from H30 in targeting younger patients and seeming to have less tendency to be isolated from blood; this, however, requires further investigation. In addition to being FQ-R, ST1193 isolates are often co-resistant to trimethoprim-sulfamethoxazole and tetracycline, but currently (unlike H30) remain susceptible to most β-lactam antibiotics. Discovery of the basis for the global expansion of ST1193 could provide insights into how successful clonal groups of multidrug-resistant E. coli emerge and what interventions could limit their spread.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Professor Steve Moseley and Dr Dagmara Kisiela for scientific advice and critical proofreading of the manuscript; clinical microbiological laboratories personnel in Seattle, Los Angeles, New York, and Minneapolis for assistance in collection of isolates; and Matthew Radey, Huxley Smart, Hovhaness Avagyan, Inga Avagyan, and Thalia Solyanik for technical assistance at the University of Washington laboratory and ID Genomics.

Financial support. This work was supported by National Institutes of Health (grant numbers R01AI106007 to E. V. S., and R41AI116114 and R42 AI116114-02 to ID Genomics and E. V. S.); the Office of Research and Development, Medical Research Service, Department of Veterans Affairs (grant number 1 I01 CX000192 01 to J. R. J.); the Partnership for Innovation Award, Group Health Foundation, and Development Fund Award, Group Health Research Institute; and the Gerber Foundation (to E. V. S.).

Potential conflicts of interest. J. R. J. has received research funding from Achaogen, Allergan, Merck, Syntiron, and Tetraphase; is a consultant for Janssen/Crucell and Syntiron; and has patent applications for tests to detect E. coli strains. V. L. T. and E. V. S. have patent applications to detect E. coli strains. E. V. S. is a major shareholder in ID Genomics. B. S. has received grants and nonfinancial support from Nabriva, Pfizer, Cempra, Bayer, Forge, Shionogi, Alexion, Synthetic Biologics, Paratek, Ovagene, Bioversys, Acurx, Motif, BioAIM, Synthetic Biologics, Mycomed, and ExBaq. F. C. F. has received grants, personal fees, and nonfinancial support from Cepheid and BioFire; grants and nonfinancial support from ELITech; nonfinancial support from Luminex; and personal fees from the Infectious Diseases Society of America. All other authors report no potential conflicts. All authors have 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.