Extended-spectrum beta-lactamase-producing strains among diarrhoeagenic Escherichia coli—prospective traveller study with literature review

Abstract Background Antibiotics are no longer the primary approach for treating all travellers’ diarrhoea (TD): most cases resolve without antibiotics and using them predisposes to colonization by multidrug-resistant bacteria. Data are accumulating on increasing resistance among TD pathogens, yet research into the most common agents, diarrhoeagenic Escherichia coli (DEC), remains limited. Methods A total of 413 travellers to the (sub)tropics were analyzed for travel-acquired diarrhoeal pathogens and ESBL-PE. To identify ESBL-producing DEC, ESBL-producing E. coli (ESBL-EC) isolates were subjected to multiplex qPCR for various DEC pathotypes: enteroaggregative (EAEC), enteropathogenic (EPEC), enterotoxigenic (ETEC), enteroinvasive (EIEC) and enterohaemorrhagic (EHEC) E. coli. For a literature review, we screened studies among travellers and locals in low- and middle-income countries (LMICs) on the frequency of ESBL-producing DEC, and among travellers, also DEC with resistance to ciprofloxacin, azithromycin, and rifamycin derivatives. Results Our rate of ESBL-EC among all DEC findings was 2.7% (13/475); among EAEC 5.7% (10/175), EPEC 1.1% (2/180), ETEC 1.3% (1/80) and EHEC (0/35) or EIEC 0% (0/5). The literature search yielded three studies reporting ESBL-EC frequency and thirteen exploring resistance to TD antibiotics among travel-acquired DEC. For EAEC and ETEC, the ESBL-EC rates were 10–13% and 14–15%, resistance to fluoroquinolones 0–42% and 0–40%, azithromycin 0–29% and 0–61%, and rifaximin 0% and 0–20%. The highest rates were from the most recent collections. Proportions of ESBL-producing DEC also appear to be increasing among locals in LMICs and even carbapenemase-producing DEC were reported. Conclusion ESBL producers are no longer rare among DEC, and the overall resistance to various antibiotics is increasing. The data predict decreasing efficacy of antibiotic treatment, threatening its benefits, for disadvantages still prevail when efficacy is lost.


Introduction
Uncontrolled use of antibiotics is a major driver of the ongoing antimicrobial resistance (AMR) pandemic, which threatens global health. 1 Increasing fastest in the tropics, 1 AMR is being transported worldwide by international travellers: 20-70% of visitors to low-and middle-income countries (LMICs) carry multidrug-resistant bacteria (MDR), particularly extendedspectrum beta-lactamase-producing Enterobacteriaceae (ESBL-PE), to their home country 2 -7 and may spread them further. 2, 6 During the past decade, avoiding unnecessary antibiotic use while abroad has emerged as a means to combat travel-related global spread of AMR. In addition to the general pressure to avoid unnecessary antibiotics, 1,8 this policy is particularly encouraged by findings that antibiotic use predisposes travellers to acquisition of multidrug-resistant intestinal bacteria 2-7 -and thus contributes to the global spread of AMR, colonized travellers acting as intercontinental transporters. 7 ,9,10 Special attention has been paid to treatment of travellers' diarrhoea (TD), which ranks as the most common indication for travellers' antibiotic use 11 : 5-45% of those with TD take these drugs to alleviate their symptoms. 3 -6,12-17 As described in the literature, stand-by antibiotics for TD are prescribed at pre-travel consultations for 7-20% of European 11 ,13,15 and practically all US travellers. 12,16, 17 Recently, the rates have also decreased somewhat in the USA. 18 While antibiotics certainly retain their place in treating the most severe TD cases, their use for moderate TD has recently become topical. 19 Although compared to placebo, antibiotics shorten the disease duration by 0.7-1.5 days, 20 ,21 in most TD cases, the drugs are not necessary, since the disease usually resolves spontaneously. Anti-diarrhoeals such as loperamide offer an alternative with no impact on AMR colonization 22 ; there are no studies that prove antibiotics to be clinically superior to loperamide in treatment of mild/moderate TD. 22 In discussions concerning antibiotics for TD, 11,23-25 limited attention has been given to resistance among diarrhoeagenic Escherichia coli (DEC), the most common TD pathogens 26 ; studies have mainly examined Salmonella, Campylobacter and Shigella. 27-33 DEC include several pathotypes: enteroaggregative (EAEC), enteropathogenic (EPEC), enterotoxigenic (ETEC), enteroinvasive (EIEC), enterohaemorrhagic (EHEC) or shigatoxin-producing (STEC) E. coli. 34 The paucity of resistance studies can be explained by the challenges in detecting the various DEC: as they in culture resemble any other E. coli, identifying a specific DEC type requires additional screening by PCR or other methods. 35 Resistance has been reported among DEC in LMICs against the antibiotics currently recommended for TD treatment, but for travel-acquired DEC, the rates are only provided by a few studies. Ouyang-Latimer et al. showed already 2011 a substantial increase in MIC values for ciprofloxacin and azithromycin between 1997 and 2006-08 among both EAEC and ETEC isolates from travellers to Mexico, Guatemala and India. 36 Moreover, travel-acquired ESBL-EAEC and ESBL-ETEC have been detected. 37-39 ESBL-DEC are of special interest, since for severely ill travellers hospitalized, first-line intravenous drugs include third-generation cephalosporins (3GC) ineffective against these pathogens. 40 Emergence of MDR strains among DEC is not unexpected-a similar development has been reported for other stool bacteria such as Salmonellae. 28 Scarcity of research into travel-acquired ESBL-DEC prompted us to revisit our data on 413 Finnish travellers to investigate the frequency of ESBL producers among various DEC. Since our samples were collected ten years ago and the global AMR situation is constantly deteriorating, to get a more accurate picture, we also screened the literature for investigations into ESBL producers, resistance of travel-acquired DEC to commonly used antibiotics, and rates of ESBL-DEC among locals in LMICs. Research into the resistance of TD pathogens provides fundamental information for guidance on antibiotic treatment of TD.

Materials and Methods
The first part of this two-faceted study explored the rates and geographic origin of ESBL-producing strains among DEC contracted by Finnish travellers to LMICs (Figure 1). The second part searched PubMed for original studies of DEC exploring proportions of ESBL producers (travellers and locals) and resistance to commonly used TD antibiotics (only travellers).

Study design, volunteers, samples and travel destinations
We prospectively recruited 526 Finnish travellers attending pretravel consultation at the Travel Clinic of Aava Medical Centre before their journey outside the Nordic countries for more than four nights. 3 Of these, 413 met our inclusion criteria (provided pre-and post-travel stools, filled in pre-and post-travel questionnaires, travel destination in LMICs). The details of stool collection, questionnaires and categorization of travel destinations have been described in our previous study. 3 Post-travel ESBL-producing Enterobacteriaceae (ESBL-PE) were considered as travel acquired only if pre-travel samples had been negative for ESBL-PE.
The protocol was approved by the Helsinki University Hospital ethics committee. All subjects provided written informed consent.

Collection of specimens
Briefly, faecal samples were collected before departure and from the first or second stools passed after returning home. For collection, we used swabs in Copan M40 Transystem tubes (Copan Diagnostics, Brescia, Italy). Once the samples arrived, total nucleic acids were extracted using the standard semiautomated protocol of easyMAG (bioMérieux, Marcy l'Etoile, France) and the stools were cultured (see below).

Analysis of DEC by qPCR
To explore the proportion of ESBL producers among various DEC (Figure 1), we first explored the total rates of stool samples positive for DEC by a multiplex qPCR assay, which identifies nine bacterial pathogens: Salmonella, Yersinia, Campylobacter, Vibrio cholerae, Shigella/EIEC, EHEC, ETEC, EAEC and EPEC. 44 Second, to identify ESBL-DEC in the same samples, we subjected the ESBL-EC isolates to the multiplex qPCR for DEC.

Participants
Demographics of the 13 with travel-acquired ESBL-DEC are provided in Table 1. Of them, 12/13 (92%) had TD and 2/12 (17%) took antibiotics for it. The entire study cohort's demographics have been published earlier 3 ; 67% had TD, 12% took antibiotics for it and 21% (90/430) were colonized by travelacquired ESBL-PE (none of the travellers had ESBL-DEC in their pre-travel stools).
Eight of the 13 participants with ESBL-DEC (61.5%) had travelled to South Asia, and three (23.1%) to the Southeast Asia. None of the visitors to sub-Saharan Africa or Latin America had ESBL-DEC.
Two volunteers had taken antibiotics (ciprofloxacin) for TD; both had an ESBL-DEC co-resistant to ciprofloxacin and tobramycin, whereas among those without antibiotic use, only one strain (1/11; 9.1%) was co-resistant to ciprofloxacin (Supplementary Table 1).
Literature on resistance among DEC, special focus on rates of ESBL-DEC In our literature search for studies of ESBL-DEC, we omitted those not reporting total number of DEC 43 ,45 or strain-specific travel data 46 ,47 ; these reports prove existence of ESBL-DEC, though. Instead, we selected, in accord with our initial aim, papers providing prevalence data on resistance among travelacquired DEC or rates of ESBL-DEC among DEC originating in LMICs. Due to meagre search results especially among travellers, we also reviewed our own files on TD studies.
Our search only yielded 24 original studies of ESBL-DEC rates among one or more types of DEC, three traveller studies 37-39 (Table 3), and 21 looking at locals in LMICs 48 -68 ( Table 4). As for travellers, we found four other investigations into resistance rates to 3GC. 36,69-71 In total, 13 traveller studies provided resistance rates to one or more TD antibiotics, 30,36-39,69-76 all presented below by DEC pathotype.

Resistance among EHEC/STEC strains
None of the traveller studies reviewed provided rates of antibiotic resistance for EHEC/STEC isolates.
Amaya et al. did not find any ESBL-EC among eight EHEC strains from Nicaraguan children with diarrhoea 49 (Table 4).

Resistance among EIEC strains
Our search yielded two traveller studies of resistance looking at EIEC isolates: among samples from US military in Thailand 2013-17 no resistance was detected 71 but in Nepal 2012-14, 10% of the EIEC strains proved resistant to ciprofloxacin and 30% to azithromycin. 76 In LMICs, studies among local children with diarrhoea have found the few EIEC strains to be mostly ESBL producers. 49,57,60,62,68

Discussion
Despite the vast discussion around antibiotic use for treating TD, paradoxically scant attention has been paid to resistance among the most common TD pathogens, DEC. The handful of reports published mostly do not focus on travellers. Apart from resistance to individual antibiotics, multidrug resistance is increasingly common among intestinal bacteria in clinical samples worldwide, ESBL-PE ranking as the most prevalent MDR type. 77 -79 Our data together with those from a literature search for studies among travellers and locals in LMICs destinations show an emergence of ESBL producers among DEC.

Rates of ESBL producers among DEC
Our rate, 3-7% of ESBL producers among the various DEC strains collected 2009-10, appears consistent with the three other traveller studies of ESBL-DEC: among Spanish travellers, the rates of ESBL-EAEC were 10% in 2005-6, 37 and 12.8% in 2011-17. 39 Among residents and travellers with acute diarrhoea in Kathmandu an increase from 1.5 to 35% was observed between   Table 4. 2008 and 2016. 38 These data suggest increasing rates of ESBL producers among DEC. We found more investigations into the ESBL-production of DEC among locals in LMICs than among travellers, with rates of positive findings varying by pathotype, time and destination between 0 and 80%. 48-68 It should be noted that none of the analyses focused on the main tourist destinations in Southeast Asia, Africa, or South and Central America, and the Caribbean. In 18 of the 21 studies, the data were from local children with or without diarrhoea, 48-50,52,53,55-62,64-68 highlighting the clinical concern related to resistance. Likewise, among locals, the highest rates were recorded over the most recent years, according with the steady global increase in the rates of ESBL-producing strains among all E. coli in clinical samples. 77- 79 Our search did not focus on carbapenemase-producing DEC, but we found 16 studies from LMICs reporting resistance rates of 0-50% to carbapenems among DEC. 49-68 Our samples showed no carbapenemase-producing genes. 3

ESBL producers among various DEC
In our data, the ESBL-EC rates appeared higher among EAEC than EPEC and ETEC (5.7% versus 1.1% versus 1.3%). This accords with other traveller studies reporting ESBL-EAEC rates of 10% 37 and 12.8% 39 among travellers yet amounting to 85% for locals in Iran 57 and 56% in China. 54 Likewise, substantial rates (53 and 57%) of ESBL-EAEC have been reported among clinical EAEC isolates in England; yet they do not report which of the strains were travel-acquired nor their countries origin. 46,47 For ESBL-ETEC, our rate, 1.3% (Table 2), was much lower than that found among Spanish travellers (14%) 39 or in Nepal (15%). 38 The top rates (75%) for non-travellers have been recorded among Iranian children. 57 As for EPEC, we only identified two ESBL-EPEC strains (1.1%). None of the traveller studies reviewed covered ESBL-EPEC, but among locals rates as high as 80% have been reported in Iran, 58 and 59% in Pakistan. 55 We detected no ESBL-EC among EIEC and EHEC, neither did we find in the literature any other traveller studies exploring ESBL-EC of these pathotypes; only few investigations among locals report ESBL-EC for EIEC or EHEC. 49,57,60,62, 68 We found no more than two studies looking at the rates from the other angle, describing the rates of a given pathotype among travel-acquired ESBL-DEC: rates of 14% in 2009-10 43 and 57% in 2017-18 45 have been shown for ESBL-DEC.

Geographic distribution of ESBL-DEC
Most of our ESBL-DEC originated in South Asia, which also proved to have the highest rates of ESBL-DEC among DEC: 18.2% of EAEC strains were ESBL producers. Indeed, South Asia also has exceptionally high-resistance rates among gram-negative bacteria in clinical samples 80, 81 and top ESBL-PE colonization rates among visitors. 2-6,82,83 Our data agree with previous data showing higher resistance rates among EAEC strains from South or Southeast Asia (33.3%; 4/12) than those from Africa (6.3%; 1/16) and Latin America (0%; 0/11). 39 Resistance to commonly used TD antibiotics While our own results centre around ESBL-DEC, we also reviewed the literature for data on resistance among travelacquired DEC to commonly used TD antibiotics (fluoroquinolones, azithromycin and rifaximin). Recent traveller studies 39,76 present alarming data: for EAEC strains resistance rates of 15-42%, 33-61% and 0% to fluoroquinolones, azithromycin and rifaximin, and for ETEC 23-33%, 22-29% and 0%, respectively.

Resistance genes among DEC
Our data include thirteen ESBL-DEC isolates, with bla CTX-M-1 as the most common finding in genetic analyses , followed by bla TEM . Only a small proportion of our strains carried the

Clinical implications
While ESBL-EC are considered resistant to 3GC (e.g. ceftriaxone), the resistance profile as such does not cover the most commonly used TD regimens, i.e. fluoroquinolones, azithromycin and rifaximin. Unfortunately, however, ESBL-producing strains often harbour co-resistance to other antibiotics, especially fluoroquinolones. 85,86 Of our ESBL-DEC strains, 3/13 (23.1%) were co-resistant to fluoroquinolones, yet higher co-resistance rates have been reported among travel-acquired ESBL-PE in general, particularly for the South Asia 2 ,5,42,45,82 and related to fluoroquinolone intake abroad. 42 Indeed, ESBL-producing strains are of special concern, since in cases severe enough to require hospitalization empiric treatment often relies on either 3GC or fluoroquinolones. 40 Interpreting the efficacy of various antibiotics is somewhat complicated, for faecal antibiotic levels tend to exceed the minimum inhibitory concentration (MIC). 23 Furthermore, presence of antibiotics in stools, while indicating an antibiotic pressure to other intestinal bacteria, may also drive transfer of resistance genes to other Enterobacteriaceae, some of which are potential pathogens. 85 ,87 An ineffective drug does not offer benefits, and yet retains its disadvantages. Although the adverse effects rate appears to be low, 88 recently, for example, the US Food and Drug Administration has warned about some serious adverse effects of fluoroquinolones (e.g. tendinitis and prolonged QT interval) and azithromycin (e.g. prolonged QT interval), 89,90 the most popular TD antibiotics. Furthermore, data are lacking on the suggested smaller impact of one-day antibiotic treatment on acquisition of MDR bacteria abroad. The adverse effect profile would favour rifamycins such as rifaximin. However, the drug is nonabsorbable and should not be used in cases with fever and invasive disease-i.e. it does not meet the most important indications for antibiotics. We only found a few studies exploring resistance rates to rifaximin among DEC; Ouyang-Latimer et al. 36 reported 16-25% resistance rates among ETEC already in 2011.

Limitations of our data
Firstly, collected 2009-10, our strains do not fully represent the current situation. Unfortunately, though, the same applies to the other traveller studies found in our search, only three of which provide data from a later time period. 38,39, 76 The increase in resistance recorded among locals suggests growing pressure also for travellers. Our data may thus present a slight underestimation, calling for updated surveillance.
Secondly, qPCR of stools cannot distinguish whether the samples contain one DEC strain or several of similar type. Likewise, in culturing ESBL-EC strains, those which appear phenotypically different are picked, and therefore strains may be missed that are similar or of only a slightly different phenotype, but genetically unlike. Fortunately, these sources of error may at least partly overcome one another.
Thirdly, in the various studies reviewed there are methodological differences (assessment of the various DEC, pre-analytical handling of the specimens, etc.); therefore, the data may not be fully comparable.

Conclusions
ESBL-producing DEC are no longer rare, particularly in Asia. Among travel-acquired DEC, their rates appear fairly low as yet, but in many regions, increase is already seen among DEC isolated from locals with acute diarrhoea, also portending increase among travel-acquired DEC, many strains even to be carried by travellers to their countries. While antibiotics certainly retain their place in the treatment of the most severe TD cases, data showing increasing resistance among stool pathogens further encourage cutting back on use of antibiotics for TD, and opting for non-antibiotic alternatives for mild and moderate cases. After all, an ineffective drug, while obviously useless, retains all its disadvantages.