Trends in Population Dynamics of Escherichia coli Sequence Type 131, Calgary, Alberta, Canada, 2006–2016

Global expansion of antimicrobial drug–resistant Escherichia coli sequence type (ST) 131 is unrivaled among human bacteria. Understanding trends among ST131 clades will help with designing prevention strategies. We screened E. coli from blood samples (n = 1,784) obtained in Calgary, Alberta, Canada, during 2006, 2012, and 2016 by PCR for ST131 and positive samples (n = 344) underwent whole-genome sequencing. The incidence rate per 100,000 residents increased from 4.91 during 2006 to 12.35 during 2012 and 10.12 during 2016. ST131 belonged to clades A (10%), B (9%), and C (81%). Clades C1-nonM27 and B were common during 2006, and C2 containing blaCTX-M-15, C1-M27 containing blaCTX-M-27, and A were responsible for the increase of ST131 during 2012 and 2016. C2 was the most antimicrobial drug–resistant subclade and increased exponentially over time. Eradicating ST131, more specifically the C2 subclade, will lead to considerable public health benefits for persons in Calgary.

Global expansion of antimicrobial drug-resistant Escherichia coli sequence type (ST) 131 is unrivaled among human bacteria. Understanding trends among ST131 clades will help with designing prevention strategies. We screened E. coli from blood samples (n = 1,784) ob-  27 , and A were responsible for the increase of ST131 during 2012 and 2016. C2 was the most antimicrobial drug-resistant subclade and increased exponentially over time. Eradicating ST131, more specifically the C2 subclade, will lead to considerable public health benefits for persons in Calgary.

Study Population
We conducted a retrospective cohort study in Calgary by using all E. coli human clinical isolates from blood cultures processed by a centralized laboratory system (Alberta Precision Laboratories) during 2006, 2012, and 2016. All blood culture samples from adults and children in inpatient and outpatient settings were included.

Clinical Data
Clinical information corresponding to source patients at the time of the E. coli bloodstream infection was obtained by using Sunrise Clinical Manager (Allscripts Healthcare Solutions, Inc., https://www.allscripts. com). A case-patient with an E. coli bloodstream infection was defined as a patient with systemic inflammatory response and documented growth of an E. coli isolate in a blood culture. Incident case-patients were defined as Calgary residents with a first isolation of E. coli from blood. Repeat E. coli from blood were excluded. Bloodstream infections were defined as community acquired, hospital acquired, or healthcare associated (11).

Bacterial Isolates, Identification, and Susceptibility Testing
All E. coli isolates from blood were routinely stored at Alberta Precision Laboratories and available for this study. Unique isolates recovered during January 1-December 31, 2006, 2012, and 2016 were obtained from the frozen depository.

Statistical Analysis
The Fisher exact test was used to perform pairwise comparisons of factors between clades, t-test was used for age comparisons, and p values obtained within individual categories were adjusted for multiple comparisons by using the false discovery rate (24). Population data were extracted from census reports from Statistics Canada (https://www.statcan.gc.ca) and used to estimate incidence rates (IRs) on the basis of a Poisson distribution. The Mann-Whitney test was used to compare antimicrobial resistance and virulence scores between clades. The effect of eliminating subclade C2 on nonsusceptibility and IRs was assessed by using Fisher exact and Poisson tests, for which population characteristics were compared with the presence and absence of subclade C2 isolates. The p values were adjusted for multiple comparisons accordingly. All analyses were conducted in R version 3.6.1 (25). Statistical significance was set at the 5% level.

Incidence Rates and Population Dynamics of ST131 Clades
The IR per 100,000 residents with ST131 bloodstream infections in Calgary increased from 4.91  (Figure).
Clade A was absent among ST131 during 2006 and then increased to 12% of the ST131 population during 2012 and 11.3% of the ST131 population during 2016 (p<0.01 for both comparisons) ( Figure). The IR of clade A increased from 0 to 1.48/100,001 residents during 2012 and to 1.15/100,000 residents during 2016 (p<0.001 for both comparisons) ( Table 2). B was the second most common clade during 2006 (26.4% of the total ST131 population), but decreased to 6% of the ST131 population during 2012 and to 6.4% of the ST131 population during 2016 (p<0.001 for both comparisons) ( Figure). The IR of clade B decreased from 1.30/100,000 residents to 0.74/100,000 residents during 2012 and to 0.65/100,000 residents during 2016 (Table 2).

Removal of Subclade C2
Eliminating subclade C2 would have decreased the incidence rate of ST131 bloodstream infections from 12.35/100,000 residents to 7.16/100,000 residents during 2012 and from 10.12/100,001 residents to 5.31/100,000 residents during 2016 (p<0.001 both comparisons). In addition, eliminating subclade C2 would have resulted in a significant reduction of not susceptible rates for amoxicillin/clavulanic acid, ciprofloxacin, ceftriaxone, and tobramycin for ST131 causing bloodstream infections in Calgary (2006, 2012, and 2016) (p<0.05 for all comparisons).

Quinolone Resistance-Determining Regions and Antimicrobial Resistance Determinants
The combination of mutations in gyrase A genes (gyrA S83L and gyrA D87N) and DNA topoisomerase IV genes (parC S80I, parC E84V, and parE I529L) in the quinolone resistance-determining regions were present in all C1 and C2 isolates (Table 3). Nearly all (97%) ST131 isolates contained the parE I529L mutation. Most (85%) clade A isolates had the gyrA S83L mutation; for 5 isolates, this mutation was combined with gyrA D87N and parC S80I, and 1 isolate had the gyrA S83L, gyrA D87N, parC S80I, and parC E84V combination. Mutations in gyrA and parC were rare in clade B; 2/32 isolates had the gyrA S83L and parC S80I mutation combination (Table  3). One subclade C0 isolate had only the gyrA S83L mutation, and another C0 isolate had the gyrA S83L, gyrA D87N, parC S80I, and parC E84V combination.

Discussion
The abrupt global expansion of ST131 during the 2000s is unrivaled among human bacteria and is a real-world model for the evolution of antimicrobial-resistant high-risk clones (10). This study describes the clinical features, incidence rates, genomic characteristics, and changes in population structure of ST131 clades causing bloodstream infections in a large centralized region of Canada over an 11-year period (2006-2016). The incidence rates and prevalence of ST131 increased over the time period, mostly caused by an influx of subclades C2 with bla CTX-M-15 and C1-M27 with bla CTX-M-27 . Such results reinforce the possible role of CTX-M enzymes in the evolutionary success of ST131 (10). The presence of bla CTX-M-14 among C1-nonM27 isolates did not provide a beneficial advantage to this subclade. This finding is probably caused by clonal interference among 2 clones that have acquired different beneficial mutations competing in the same environment (27).
The population structure of ST131 in the Calgary region was dominated by clade C, which is similar to results from a previous large global study (28). The C clade originated from clade B during the mid to late 1980s by acquisition of several prophages, genomic islands, the fimH30 allele, and mutations within gyrA and parC that likely transpired in North America (29,30). The C clade in this study was mostly responsible for healthcare-associated urinary tract infections. C2 was the most common and most antimicrobial-resistant subclade in this collection and was associated with group 2 plasmids, bla CTX-M-15 and aac(6')-Ib-cr, as well as the virulence factors iha, hlyA, and cnf1. This subclade became prominent during 2012 and 2016 and showed the highest IRs among all subclades during this period. The increase of C2 correlated with the presence of CTX-M-15. Elimination of the C2 subclade through vaccination or phagetherapy programs (31), will lead to major decreases in incidence and antimicrobial-resistant rates among ST131 causing bloodstream infections in Calgary.
The C1-nonM27 subclade was the most common subclade during 2006 and associated with group 1 plasmids, bla CTX-M-14 , and aac (3)-IId. Overall, the C1-M27 subclade was rare (especially during 2006 and 2012) but increased substantially during 2016, which correlated with the presence of bla CTX-M-27 . The C1-M27 subclade has previously been responsible for increases in ESBL-producing E. coli from Japan and was also present among ST131 obtained from Thailand, Australia, Canada, and the United States (6). The ST131 C1-M27 subclade is currently emerging in Germany (32) and France (33) and is responsible for 27% of 144 clinical ST131 obtained from different sites in Europe (34).
Clade A is likely the ancestral lineage of ST131 and probably originated in Southeast Asia during the mid to late 1880s (30). Clade A isolates are generally sensitive to antimicrobial drugs and appear to occupy distinct ecologic niches, such as waste water (35). Results from this study show that clade A isolates have high not susceptible rates for trimethoprim/sulfamethoxazole, gentamicin, and tobramycin and were associated with community-associated and healthcare-associated urinary tract infections in younger patients. The virulence factors afaABCD, draABCDP, vat, and traT were common in clade A. Also, clade A was absent among ST131 from 2006 but became the third most common clade during 2012 and 2016, replacing clades B and C0 during these periods.
Clade B emerged from clade A in the early 1900s and most likely occurred in North America (10,30). Members of clade B are antimicrobial susceptible, and several intermediate subclades have been identified (29). Our study showed that clade B isolates were the second most common clade during 2006 but decreased substantially during 2016. This clade was the most antimicrobial sensitive ST131 clade in Calgary and was associated with community-acquired urinary tract infections and virulence factors afaABCD, draABCDP, kpsMII, and ibeABC.
Previous data have shown that gyrA S83L mutations occurred first among fluoroquinolone-resistant  E. coli and is a major initial step for establishing relative fitness among antimicrobial-resistant isolates (36). Our study showed that gyrA mutations were rare among clade B isolates, but parE I529L mutations were common. This finding suggests that parE I529L mutations are the first to occur among fluoroquinolone-resistant ST131. The order in which these mutations arise might play a major role in establishing fitness in ST131 (37).
Our study had some limitations. Only patients in Calgary who had positive blood cultures for E. coli were included, which excluded those with E. coli bloodstream infections from whom no blood samples were submitted for culture. Therefore, incidence rates should be considered as conservative estimates of ST131 bloodstream infections in Calgary, especially for patients infected with clades A and B, who tended to be younger (i.e., clade A infections) and from the community (i.e., clade B infections). Such patients were less likely to have had blood cultures taken than patients who are older or who had previous contact with the healthcare system.
The novel approach for our study used population-based surveillance to describe the incidence rates, specific characteristics, and trends among ST131 clades over an 11-year period in a well-defined human population. We showed major differences in IRs, frequencies, resistance patterns, antimicrobial resistance determinants, grouped plasmid types, virulence factors, and trends over time for different clades. We provided insights into the evolution of ST131 clades in a large well-defined region of Canada. The population structure of ST131 in large geographic healthcare regions is dynamic and has continuous interplay between different subclades.
A previous study showed that eliminating ST131 would substantially decrease the overall IR and antimicrobial-resistant burden within E. coli causing bloodstream infections in the Calgary region (7). This study identified ST131 subclade C2 as the predominant and most antimicrobial-resistant subclade in Calgary, which is increasing exponentially over time. Eradicating ST131, more specifically the C2 subclade, will lead to considerable public health benefits for persons in Calgary.