Community carriage of ESBL-producing Escherichia coli and Klebsiella pneumoniae: a cross-sectional study of risk factors and comparative genomics of carriage and clinical isolates

ABSTRACT The global prevalence of infections caused by extended-spectrum β-lactamase-producing Enterobacterales (ESBL-E) is increasing, and for Escherichia coli, observations indicate that this is partly driven by community-onset cases. The ESBL-E population structure in the community is scarcely described, and data on risk factors for carriage are conflicting. Here, we report the prevalence and population structure of fecal ESBL-producing E. coli and Klebsiella pneumoniae (ESBL-Ec/Kp) in a general adult population, examine risk factors, and compare carriage isolates with contemporary clinical isolates. Fecal samples obtained from 4,999 participants (54% women) ≥40 years in the seventh survey of the population-based Tromsø Study, Norway (2015, 2016), were screened for ESBL-Ec/Kp. In addition, we included 118 ESBL-Ec clinical isolates from the Norwegian surveillance program in 2014. All isolates were whole-genome sequenced. Risk factors associated with carriage were analyzed using multivariable logistic regression. ESBL-Ec gastrointestinal carriage prevalence was 3.3% [95% confidence interval (CI) 2.8%–3.9%, no sex difference] and 0.08% (0.02%–0.20%) for ESBL-Kp. For ESBL-Ec, travel to Asia was the only independent risk factor (adjusted odds ratio 3.46, 95% CI 2.18–5.49). E. coli ST131 was most prevalent in both collections. However, the ST131 proportion was significantly lower in carriage (24%) versus clinical isolates (58%, P < 0.001). Carriage isolates were genetically more diverse with a higher proportion of phylogroup A (26%) than clinical isolates (5%, P < 0.001), indicating that ESBL gene acquisition occurs in a variety of E. coli lineages colonizing the gut. STs commonly related to extraintestinal infections were more frequent in clinical isolates also carrying a higher prevalence of antimicrobial resistance, which could indicate clone-associated pathogenicity. IMPORTANCE ESBL-Ec and ESBL-Kp are major pathogens in the global burden of antimicrobial resistance. However, there is a gap in knowledge concerning the bacterial population structure of human ESBL-Ec/Kp carriage isolates in the community. We have examined ESBL-Ec/Kp isolates from a population-based study and compared these to contemporary clinical isolates. The large genetic diversity of carriage isolates indicates frequent ESBL gene acquisition, while those causing invasive infections are more clone dependent and associated with a higher prevalence of antibiotic resistance. The knowledge of factors associated with ESBL carriage helps to identify patients at risk to combat the spread of resistant bacteria within the healthcare system. Particularly, previous travel to Asia stands out as a major risk factor for carriage and should be considered in selecting empirical antibiotic treatment in critically ill patients.

IMPORTANCE ESBL-Ec and ESBL-Kp are major pathogens in the global burden of antimicrobial resistance. However, there is a gap in knowledge concerning the bacte rial population structure of human ESBL-Ec/Kp carriage isolates in the community. We have examined ESBL-Ec/Kp isolates from a population-based study and compared these to contemporary clinical isolates. The large genetic diversity of carriage isolates indicates frequent ESBL gene acquisition, while those causing invasive infections are more clone dependent and associated with a higher prevalence of antibiotic resistance. The knowledge of factors associated with ESBL carriage helps to identify patients at risk to combat the spread of resistant bacteria within the healthcare system. Particularly, previous travel to Asia stands out as a major risk factor for carriage and should be considered in selecting empirical antibiotic treatment in critically ill patients.

Study population and design
Our study sample was drawn from Tromsø7, the last of seven cross-sectional health surveys conducted between 1974 and 2016 in Tromsø municipality, Norway (https:// uit.no/research/tromsostudy). Tromsø is representative of a Northern European, urban population (39). Tromsø7 (March 2015-October 2016) included questionnaires and two clinical visits (https://uit.no/research/tromsostudy/project?pid=708909). Unique national identity numbers from the official population registry were used to invite all citizens ≥40 years (n = 32,591). Sixty-five percent (n = 21,083, 11,074 women) attended the first clinical visit in the study (Fig. 1).
A selection of 9,320 persons attending the first visit was invited for a second visit, also including 3,154 former Tromsø Study participants not already included in the random selection process, which was required for other clinical research purposes. From March 2015 to March 2016, 5,800 participants at the first visit were consecutively provided a fecal self-sampling kit. Participants collected fecal material using nylon-flocked ESwab 490CE.A (Copan, Brescia, Italy). In total, 87% (n = 5,042) returned a sample either at the second visit or by mail to the laboratory.
All 5,042 fecal samples were screened for the presence of ESBL-producing Ec and Kp via selective culture (see below). All participants completed two self-administered structured questionnaires on sociodemographics, smoking, alcohol use, hospitalization, drug use, and travel abroad. We excluded 13 participants with the wrong or missing sample identification numbers, two retracting consent to medical research, and 28 with incomplete questionnaires for a final study population of 4,999 participants ( Fig. 1). We analyzed the association between ESBL-Ec gastrointestinal carriage and different risk factors in 4,999 participants ( Fig. 1). Next, we conducted a sensitivity analysis studying the association between Kp gastrointestinal carriage and ESBL-Ec carriage among 2,973 participants additionally screened for Kp in our previous study (25), irrespective of resistance ( Fig. 1).

Isolation of ESBL-producing E. coli and K. pneumoniae
We added 200 µL of 85% glycerol to the ESwab tubes on arrival at the local microbiologi cal laboratory and stored the samples at −80°C. From the thawed media, 100 µL were plated onto CHROMagar ESBL (CHROMagar, Paris, France) and incubated for 48 hours at 37°C. Pink, purple, and blue colonies suspected of being ESBL-producing Ec or Klebsiella spp. were identified using mass spectrometry (matrix-assisted laser desorption ioniza tion-time of flight [MALDI-TOF]; Bruker Daltonics, Bremen, Germany). The first colony identified as either E. coli, K. pneumoniae, or Klebsiella variicola from each sample was kept and further analyzed. All samples were plated on cysteine lactose electrolyte deficient agar (MAST Group, Bootle, UK) to assess the growth of fecal flora and validity of the samples.

K. pneumoniae isolation
The screening strategy and isolation procedure for Kp gastrointestinal carriage of 2,973 participants in Tromsø7 are described in detail elsewhere (25). Briefly, we plated and screened the fecal samples onto the selective SCAI (Simons citrate agar with inositol; both Sigma-Aldrich, Darmstadt, Germany) medium and identified suspected colonies using MALDI-TOF.
The clinical ESBL-producing Ec included 118 isolates out of 123 representing all ESBL-Ec isolates collected in 2014 (55), as part of the yearly Norwegian Surveillance Program of Antimicrobial Resistance (NORM). The surveillance program included all clinical microbiological laboratories in Norway. The sampling period was 6 months and 2 days for blood and urine isolates, respectively. Genome sequencing of five isolates was unsuccessful. Before sequencing, the NORM 2014 isolates were stored at −80°C and then sent to GATC Biotech AG (part of Eurofins Genomics/Eurofins Scientific) in Germany for DNA isolation and WGS. Raw reads were trimmed using Trimmomatic v0.39 and assembled with SPAdes v3.15.0 (56,57). Contigs shorter than 200 bp were discarded. The genomic data were analyzed as described above.

Statistical analysis
Our primary analysis was a multivariable logistic regression model, with the outcome variable ESBL-Ec gastrointestinal carriage using SPSS v.26.0 (SPSS, Inc., Chicago, IL, USA). We analyzed factors associated with ESBL-Ec gastrointestinal carriage among 4,999 participants (Table 1). Both the primary (Table 1) and the sensitivity analysis (Table  S2) were multivariable logistic regression analyses. Explanatory variables were selected with the help of a directed acyclic graph constructed using DAGitty v3.0 ( Fig. S3 and S4) (64). All explanatory variables were kept in the fully adjusted models. Multicollinear ity between the entered variables was assessed by calculating the variance inflation factor (VIF) and tolerance statistic. Multicollinearity was not a problem with VIF >10 and tolerance statistic <0.2 (65). The strength of the associations was examined by calculat ing adjusted odds ratios (AORs) with 95% confidence interval (CI). Two-sided P-values <0.05 were considered statistically significant. The prevalence of ST131 among carrier and clinical isolates was compared by calculating the OR with 95% CI using logistic regression in SPSS. The comparison of ESBL-Ec and phenotypic resistance proportions were assessed using χ 2 test.

RESULTS
We detected gastrointestinal carriage of putative ESBL-Ec/Kp in 188 of 4,999 randomly selected participants who provided fecal samples in the seventh survey of the popula tion-based Tromsø study (Tromsø7) (39) (Table S1; Fig. 2). Overall, 87% of participants receiving a sampling kit returned a fecal sample. In total, 180 Ec and 9 Kp putative ESBL-positive isolates were isolated from the 188 ESBL screening-positive fecal samples. Both ESBL-Ec and ESBL-Kp were detected in one sample.
Among the nine screening ESBL-positive Kp isolates, one harbored bla CMY-2 and four did not express an ESBL phenotype. Consequently, four samples were considered positive for ESBL-producing Kp, corresponding to a prevalence of 0.08% (0.02%-0.20%), all among male participants.

Factors associated with gastrointestinal carriage of ESBL-E. coli
We analyzed data from 4,999 participants (Table S1). Median age was 65 years (interquar tile range 58-70 years, no sex difference) of the whole study population. In multivariable logistic regression analyses adjusted for all explanatory variables, only travel to Asia in the past 12 months was associated with ESBL-Ec gastrointestinal carriage with an AOR of 3.46 (2.18-5.49) ( Table 1). Among participants reporting hospitalization in the past year, or recent use of antibiotics or acid-suppressive medication, we observed a non-significant increase in the prevalence of ESBL-Ec.
Considering the overlapping risk factors between ESBL-E and Kp gastrointestinal carriage identified in the literature (15,22,26,27) and in our previous study (25), we conducted a sensitivity analysis with Kp carriage as a risk factor in a subgroup of 2,973 participants (of the total 4,999 participants in this current study) previously screened for Kp (Table S2). In this model including Kp carriage, AOR was 1.65 (0.98-2.77), indicating a possible association between Kp and ESBL-Ec carriage (Table S2). However, no significant differences in the estimates of the risk factors were observed compared to the model without Kp carriage conducted on the whole study population (Table 1; Table S2), indicating the validity of our primary model.
Using an SNP cutoff of ≤17 (60), we detected 2 putative clusters (Table S5) among 5 of 284 isolates. All five were carriage isolates. One ST357 cluster (6-9 SNP differences) consisted of three isolates and the other cluster of two ST131 isolates (4 SNP differences). We detected no clusters among the clinical isolates.

DISCUSSION
Our study contributes to the knowledge of prevalence of, and factors associated with, ESBL-Ec/Kp gastrointestinal carriage in a general adult population and the bacterial population structure of carriage isolates. The comparison to a contemporary collection of clinical ESBL-Ec isolates revealed differences in the population structure and the prevalence of phenotypic resistance between carriage and clinical isolates. Travel to Asia was identified as a major risk for ESBL-Ec gastrointestinal carriage.
An ESBL-Ec carriage prevalence of 3.3% (2.8%-3.9%) is lower but comparable to previous community-based data from Europe including Sweden 4.4% (3.5%-5.3%, n = 2,134; data collected 2012-2013) (13), the Netherlands 4.5% (3.9%-5.1%, n = 4,177; 2014-2016) (14), and a Norwegian study 4.9% (2.7%-8.1%, n = 284; 2014-2016) (17) using similar screening approaches. We identified significant differences in the ESBL-Ec population structure between the community and clinical isolates. The globally disseminated phylogroup B2 clone ST131 has been identified as a key contributor to the increase in ESBL prevalence (68) and in a longitudinal study of E. coli bloodstream isolates we identified ST131 to be the single largest contributor to the increase in the prevalence of ESBL-Ec in Norway (30). We also observed the predominance of ST131 in both of our collections. However, the proportion is significantly lower in carriage isolates due to lower numbers of the multidrug-resistant subclade C2. Moreover, the carriage population had a higher proportion of phylogroup A, associated with asymptomatic intestinal carriage in humans (69,70), and a significantly greater ST diversity overall, compared to the clinical isolates. These observations indicate that the acquisition of ESBL genes frequently occurs in a variety of E. coli lineages colonizing the gut. However, there are differences in the colonization potential of E. coli lineages and the risk of invasive infection by ESBL-Ec which seems to be clone dependent (68,71). The higher odds for the infection that we detected for ST131 is similar to that of the Swedish study (AOR 3.4, 1.8-6.4) indicating a higher pathogenicity potential of ST131 compared to commensal E. coli lineages of phylogroup A, such as ST10 (13). Moreover, we found that ST131 subclade A, previously reported with less resistance, and the multidrug-resistant subclade C2 had higher odds for infection, and this may contribute to the sustained establishment of these subclades among bloodstream infections in Norway (30).
Assuming a 100% colonization rate of E. coli, the large proportion of STs notorious as common causes of extraintestinal clinical infections (e.g., ST131, ST405, ST38, and ST648) (30,32,33) could at least partly explain the higher prevalence of ESBL among E. coli causing bloodstream infections in Norway (5.8% in 2016) (72) compared to the carriage prevalence of 3.3% identified here. We also observed emerging clones such as ST1193, which appears to have disseminated rapidly worldwide over the last decade (66, 67). The low prevalence of ESBL-Kp gastrointestinal carriage (0.08%) is consistent with previous community-based reports (9,14,17).
The identification of clusters with closely related isolates could indicate putative clonal spread. This could include within-household and social network transmission or nosocomial spread. However, we did not have access to epidemiological data to examine this further.
In line with other studies, we found a strong association between ESBL-Ec carriage and travel to Asian regions (13-15, 17, 22). This supports the current patient screen ing recommendations for ESBL-producing Gram-negative bacteria after a hospital stay abroad in the past year before hospital admission in Norway (73). In contrast to a Swedish and a Dutch study (15,74), we did not identify travelers' diarrhea as a risk factor for ESBL-Ec carriage. However, our study was not designed to specifically investigate international travelers but rather focused on risk factors in the general adult population.
We found no association between ESBL-Ec gut carriage and factors such as hospitali zation, antibiotic use, and acid-suppressive medication, and conflicting results have been detected in previous studies (22,75). Hospitalization as a risk factor has mainly been reported in studies investigating patients with ESBL-E infections (23,76). In line with most studies that assessed risk factors regarding ESBL-E carriage in individuals in the community, we did not identify hospitalization as an independent risk factor (13,22,74). This may be due to the increased ESBL prevalence not only in hospitals but also in the community over the last decades and implies that boundaries have become blurred between those two settings (9,22,77,78).
The non-significant effect of antibiotic use is mainly due to the limitations of the drug variable in our study which is based on self-reported data, including topical antibiotics and covers only the last 2 weeks before self-sampling. As antibiotic use has been found as a risk factor for resistance in many other studies (14,15,22), we cannot rule out that antibiotic use plays a role in ESBL-E carriage. There are reports identifying an association between the use of gastric acid-suppressive medication and intestinal colonization or infections with ESBL-E (18,21,23,26,27,76). However, a Dutch study comparable to ours did not find an association between proton pump inhibitor use and ESBL-E carriage in the overall analysis (14). Interestingly, we found a possible association between Kp and ESBL-Ec carriage. This may support the previously described link between ESBL-E, VRE (7,28), and Kp colonization among intensive care unit patients (29). These associations warrant further investigations to assess if a common set of risk factors for carriage of different clinically important pathogens can be identified.
Decolonization of ESBL-E gastrointestinal carriage has been investigated as a possible strategy to reduce the risk of infection and transmission. However, decolonization is not recommended due to insufficient evidence (79). Our findings could also be considered in designing future decolonization strategies.
An important strength of our study is the non-selective recruitment from the official population registry, and the high participation (87%) compared to 18.3% and 18.8% in comprehensive studies from the Netherlands (14) and Sweden (13), respectively. It is a limitation that we only captured the general population 40 years and older. However, other studies have not found an association between age and ESBL-E carriage (13,14). If age is associated with specific Ec STs, this might potentially bias the ST-spe cific estimates. Moreover, more extensive data on drug use would have strengthened the analyses. Additionally, the genomic diversity of carriage isolates is likely to be underestimated due to the isolation and sequencing of only one colony per fecal sample. A previous nationwide genomic study on E. coli causing bloodstream infections detected no discernible spatiotemporal spread or phylogenetic structure within Norway (30), indicating limited bias in comparing a local and national collection of isolates. The short-read data did not allow us to investigate differences in plasmid population structure between carriage and clinical isolates beyond the level of replicon types.

Conclusions
The prevalence of ESBL-Ec carriage in a general adult urban Norwegian population was low reflecting the relatively low prevalence of ESBL-Ec in clinical isolates. Travel to Asia was the only independent risk factor for ESBL-Ec carriage and should be considered in terms of screening recommendations before hospital admission. The differences in ESBL-Ec populations between carriage and clinical isolates indicating a higher risk of infection dependent on the ESBL-Ec clone support the integration of genomics in risk assessments. 7

ADDITIONAL FILES
The following material is available online. Fig. S1 (mSphere00025-23-s0001.pdf). ST131 clade distribution among ESBL-E. coli carriage isolates from Tromsø7 (n=40) and clinical isolates from NORM 2014 (n=68). Fig. S2 (mSphere00025-23-s0002.pdf). Replicon type distribution among ESBL-E. coli from carriage isolates from Tromsø7 (n=166) and clinical isolates from NORM 2014 (n=118). Fig. S3 (mSphere00025-23-s0003.pdf). Directed acyclic graph (DAG) illustrating the causal relationship between ESBL-E. coli (ESBL-Ec) gastrointestinal carriage (outcome) and relevant covariates among 4,999 participants in Tromsø7. The variable 'drug use' includes antibiotic use past 14 days and acid suppressive medication past four weeks. To set up the most plausible causal relationship between covariates and the outcome, we first searched the literature for relevant factors associated with ESBL-E. coli gastrointestinal carriage. Thereafter, the DAG guided the selection of the multivariable logistic regres sion model, which was adjusted for the minimal sufficient adjustment set comprising age, drug use, hospitalization, travel abroad, and traveler`s diarrhea (i.e. the variables constituting a confounding pathway). Controlling for the minimal sufficient adjustment set warrants that confounding paths are blocked in order to minimize bias of the causal relationship. The model was not adjusted for sex due to no statistically significant sex difference in prevalence of ESBL-E. coli carriage and because it does not constitute biasing paths after adjustment. For the variables sex and alcohol consumption, a direct effect on ESBL-E. coli carriage is not described or investigated, however they are known to affect the microbiota composition. Hence, intestinal microbiota was included as an unobserved mediator of the causal relationship. Fig. S4 (mSphere00025-23-