International Journal of Antimicrobial Agents 43 (2014) 431–437

Contents lists available at ScienceDirect

International Journal of Antimicrobial Agents journal homepage: http://www.elsevier.com/locate/ijantimicag

In vitro activity of rifaximin against clinical isolates of Escherichia coli and other enteropathogenic bacteria isolated from travellers returning to the UK夽 Katie L. Hopkins a,∗ , Shazad Mushtaq a , Judith F. Richardson b , Michel Doumith a , Elizabeth de Pinna b, Tom Cheasty b, John Wain b,c, David M. Livermore a,c, Neil Woodford a a

Antimicrobial Resistance and Healthcare Associated Infections Reference Unit, Public Health England, 61 Colindale Avenue, London NW9 5EQ, UK Gastrointestinal Bacteria Reference Unit, Public Health England, 61 Colindale Avenue, London NW9 5EQ, UK c Norwich Medical School, University of East Anglia, Norwich NR4 7TJ, UK b

a r t i c l e

i n f o

Article history: Received 10 December 2013 Accepted 24 January 2014 Keywords: Travellers’ diarrhoea Rifamycin resistance ADP-ribosyltransferases rpoB blaNDM CTX-M ␤-lactamases

a b s t r a c t Rifaximin is licensed in the EU and USA for treating travellers’ diarrhoea caused by non-invasive bacteria. Selection for resistance mechanisms of public health significance might occur if these are linked to rifamycin resistance. Rifaximin MICs were determined by agar dilution for 90 isolates each of Escherichia coli, Shigella spp., nontyphoidal Salmonella enterica, typhoidal S. enterica and Campylobacter spp., an additional 60 E. coli with CTX-M ESBLs isolated from patients with travellers’ diarrhoea, and 30 non-diarrhoeal carbapenemase-producing E. coli. Comparators were rifampicin, ciprofloxacin, azithromycin, trimethoprim/sulfamethoxazole and doxycycline. Isolates with rifaximin MICs>32 mg/L were screened for arr genes, and critical rpoB regions were sequenced. Rifaximin was active at ≤32 mg/L against 436/450 (96.9%) diverse Enterobacteriaceae, whereas 81/90 (90%) Campylobacter spp. were resistant to rifaximin at ≥128 mg/L. Rifaximin MICs were ≥128 mg/L for two Shigella and five MDR E. coli producing NDM (n = 3), OXA-48 (n = 1) or CTX-M-15 (n = 1). Two of the five MDR E. coli had plasmids harbouring arr-2 together with blaNDM , and two (one each with blaNDM and blaCTX-M-15 ) had His526Asn substitutions in RpoB. The rifamycin resistance mechanism remained undefined in one MDR E. coli isolate (with blaOXA-48 ) and the two Shigella isolates. Rifaximin showed good in vitro activity against diverse Enterobacteriaceae but was largely inactive against Campylobacter spp. Rifaximin has potential to coselect MDR E. coli in the gut flora, but much stronger associations were seen between ESBL and/or carbapenemase production and resistance to alternative treatments for travellers’ diarrhoea, notably ciprofloxacin and azithromycin. Crown Copyright © 2014 Published by Elsevier B.V. on behalf of International Society of Chemotherapy. All rights reserved.

1. Introduction Diarrhoea is the most common illness reported in international travellers, with bacterial pathogens (including enteropathogenic Escherichia coli, Salmonella enterica, Campylobacter spp. and Shigella spp.) responsible for most cases. Although travellers’ diarrhoea is usually self-limiting, ciprofloxacin or azithromycin may be used to reduce the duration and severity. In contrast to these two agents, rifaximin, a semisynthetic rifamycin licensed for the treatment of travellers’ diarrhoea caused by non-invasive bacterial pathogens

夽 These data were first presented at the 22nd European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), 31 March–2 April 2012, London, UK [abstracts P1829 and P1293]. ∗ Corresponding author. Tel.: +44 20 8327 7061; fax: +44 20 8200 7449. E-mail address: [email protected] (K.L. Hopkins).

in the EU and USA, has minimal absorption from the gut (32 mg/L for at least one rifamycin were screened by PCR for arr genes. In addition, a 2240-bp internal fragment of rpoB was amplified and sequenced, corresponding to amino acid residues 143–690 and spanning the regions associated with rifampicin resistance [4]. For primer sequences, see Table 1. Plasmids were extracted by alkaline lysis, and transfer of rifamycin resistance was attempted by electroporation with E. coli Alpha-Select Electrocompetent cells (Bioline, London, UK). Transformants were selected on Luria–Bertani agar (Oxoid Ltd., Basingstoke, UK) containing 50 mg/L rifampicin. Genes encoding NDM, CTX-M or acquired AmpC ␤-lactamases as well as the RmtC or ArmA 16S rRNA methyltransferases (conferring high-level aminoglycoside resistance) were sought in transformants by PCR as previously described [8–10]. 3. Results

2.2. Antibiotics and susceptibility testing

3.1. Susceptibility to rifamycins

Rifaximin was from Norgine Ltd. (Uxbridge, UK). Rifampicin, ciprofloxacin, azithromycin, doxycycline and trimethoprim/sulfamethoxazole (co-trimoxazole) were from Sigma

The MIC of both rifamycins for E. coli ATCC 25922 was 16 mg/L. Amongst the panel of Enterobacteriaceae isolated from travellers returning to the UK with diarrhoea, the rifaximin MICs were

Table 1 Primers designed and used in this study. Name

Primer sequence (5 → 3 )

Target

Amplicon size (bp)

arr2,3-F arr2,3-R arr4-F arr4-R arr5-F arr5-R arr6-F arr6-R arr7-F arr7-R rpoB-F rpoB-R Ecoliseq-1 Ecoliseq-2 Salmseq-1 Salmseq-2

CTATCATGGAACCAAAGCCA CAACGCCAACAATTCTCAAG ACATCTACATCGTTGAACCG TGAAGATCCTCCAGAGACG GCAATCCAACACAGTCCTAT GAGGTCTGCCAGAGAATCTA CGTGGCTACATATACATCGTC CGCTCTAGGTCCTCTAAAGACT AAAACTCGCAATAGGTGACTT GTCCTCTACTACCTCCACAATC GGTAAACGTCCACAAGTTCTG GTTCTGGTTAGAACGGGTGT GGACAACCTGTTCGTACGTA TTCGAAGCCTGCACGTTCA GAATTCGATCCGAAGGACAA TCGGGTGTACGTCTCGAACT

arr-2 and arr-3

296

arr-4

190

arr-5

119

arr-6

199

arr-7

286

rpoB, enclosing all potential mutated regions (amino acid positions 143–690)

2240

rpoB internal sequencing primer (used for Escherichia coli and Shigella spp.) rpoB internal sequencing primer (used for E. coli and Shigella spp.) rpoB internal sequencing primer (used for Salmonella enterica) rpoB internal sequencing primer (used for S. enterica)

– – – –

K.L. Hopkins et al. / International Journal of Antimicrobial Agents 43 (2014) 431–437

433

Table 2 Minimum inhibitory concentration (MIC) distributions of rifaximin and comparator agents against enteropathogenic bacteria isolated from travellers returning to the UK.

The dotted vertical lines indicate susceptible/intermediate breakpoints; the solid vertical lines indicate intermediate/resistant breakpoints; and the broken vertical lines indicate epidemiological cut-off (ECOFF) values.

unimodal, with peaks at between 4 mg/L and 32 mg/L according to the species group (highest for typhoidal salmonellae and lowest for non-typhoidal salmonellae; Table 2); MIC distributions of rifampicin for these groups similarly clustered from 8 to 32 mg/L.

Similarly, the MIC distributions of both rifaximin and rifampicin had modes of 16 mg/L for the MDR E. coli isolates, with no evidence that these were generally more resistant than the diarrhoeagenic isolates (Table 3).

434

K.L. Hopkins et al. / International Journal of Antimicrobial Agents 43 (2014) 431–437 Table 3 Minimum inhibitory concentration (MIC) distributions of rifaximin and comparator antibiotics for 90 clinical isolates of Escherichia coli that produced extended-spectrum ␤-lactamases and/or carbapenemases.

The dotted vertical lines indicate intermediate breakpoints; the solid vertical lines indicate resistant breakpoints; and the broken vertical lines indicate epidemiological cut-off (ECOFF) values. a MIC is ≤0.125 mg/L. b MIC is ≥128 mg/L.

K.L. Hopkins et al. / International Journal of Antimicrobial Agents 43 (2014) 431–437

Taking into account these MIC distributions, a provisional value of ≤32 mg/L was applied to distinguish wild-type isolates from those likely to have an acquired resistance mechanism. Based on these criteria, rifaximin was active at ≤32 mg/L against a greater proportion of the diverse Enterobacteriaceae (352/360; 97.8%) than the MDR E. coli isolates (84/90; 93.3%). This difference was not quite statistically significant (P = 0.07; http://www.graphpad.com/quickcalcs/contingency2/), whereas rifampicin was statistically more active at ≤32 mg/L against diverse Enterobacteriaceae (358/360; 99.4%) than against MDR E. coli (83/90; 92.2%) (P < 0.0001). The MIC of one or both rifamycins was >32 mg/L for two Shigella, six typhoidal salmonellae and seven MDR E. coli, with the two Shigella and six of the MDR E. coli resistant to both rifaximin and rifampicin. MICs of both rifamycins were ≥128 mg/L for the two Shigella and five MDR E. coli isolates; these E. coli variously produced NDM (n = 3), OXA-48 (n = 1) or CTX-M-15 (n = 1) enzymes. Rifaximin was less active against Campylobacter spp., with MICs ≥128 mg/L for 81/90 (90.0%) isolates, including 20 (87.0%) C. coli and 61 (91.0%) C. jejuni (Table 2); similarly, high MICs were noted for rifampicin, with values ≥128 mg/L for 82/90 (91.1%) isolates. 3.2. Susceptibility to current treatments for travellers’ diarrhoea: azithromycin and ciprofloxacin There are no clinical breakpoints for azithromycin against Enterobacteriaceae, although an ECOFF of 16 mg/L is used for Salmonella Typhi (http://bsac.org.uk/wp-content/uploads/ 2012/02/Version-12-Apr-2013 final.pdf). Applying this criterion more generally, 20/360 (5.6%) of the Enterobacteriaceae from UK travellers belonged to non-wild-type populations, with 14, mostly E. coli, showing high-level resistance (MIC≥128 mg/L) (Table 2). Azithromycin resistance (MIC>16 mg/L) was seen in a much greater proportion (40/90; 44.4%) of the MDR E. coli isolates (P < 0.0001) (Table 3). Using CLSI breakpoints [7], 27/270 (10.0%) enteropathogenic Enterobacteriaceae, excluding typhoidal salmonellae, were non-susceptible to ciprofloxacin (MICs>1 mg/L) (Table 2), as were 55/90 (61.1%) MDR E. coli isolates that produced ESBLs and/or carbapenemases (Table 3). In addition, 61/90 (67.8%) typhoidal salmonellae were resistant to ciprofloxacin at the lower breakpoint (MIC>0.06 mg/L) applied to this species. Applying the azithromycin ECOFFs for C. jejuni (wildtype ≤ 0.25 mg/L) and C. coli (wild-type ≤ 0.5 mg/L), 15/67 (22.4%) C. jejuni and 13/23 (56.5%) C. coli belonged to non-wild-type populations (Table 2). Of 90 Campylobacter isolates, 66 (73.3%) were non-susceptible to ciprofloxacin (MICs>0.5 mg/L), mostly with MICs of 8–64 mg/L. 3.3. Susceptibility to co-trimoxazole and doxycycline Using CLSI breakpoints [7], a higher proportion (76/90; 84.4%) of MDR E. coli with ESBLs and/or carbapenemases were non-susceptible to co-trimoxazole (MICs>2 mg/L) than enteropathogenic Enterobacteriaceae (141/360; 39.2%) (P < 0.0001), whilst 179/360 (49.7%) Enterobacteriaceae and 73/90 (81.1%) MDR E. coli were non-susceptible to doxycycline (MICs>4 mg/L) (P < 0.0001) (Tables 2 and 3). Applying the co-trimoxazole ECOFFs for C. jejuni (wildtype ≤ 16 mg/L) and C. coli (wild-type ≤ 4 mg/L), 1/67 (1.5%) C. jejuni and 5/23 (21.7%) C. coli belonged to non-wild-type populations (Table 2). Of 90 Campylobacter isolates, 59 (65.6%) belonged to the non-wild-type population based on the doxycycline ECOFF (wild-type ≤ 0.5 mg/L); in this case the MIC distribution was clearly bimodal, with MICs>8 mg/L for 38/59 (64.4%) isolates.

435

3.4. Rifamycin resistance mechanisms An arr-2 gene was identified in two of five MDR E. coli isolates with the highest rifamycin MICs (≥256 mg/L). The plasmids carrying arr-2 were transformed into E. coli DH5␣ and were both shown also to carry a blaNDM gene; one (Tr338) additionally carried the rmtC 16S rRNA methyltransferase gene and a CIT-type AmpC gene. MICs for the recipient and transformants are shown in Table 4. Two further MDR E. coli isolates were highly resistant to rifamycins, one with NDM-1 carbapenemase and one with CTX-M-15 ESBL; these had identical mutations (C1576A) in their rpoB genes, which were predicted to cause a His526Asn amino acid substitution. The mechanism of high-level rifamycin resistance (MIC ≥ 128 mg/L for both analogues) remained unexplained for one MDR E. coli isolate producing OXA-48 carbapenemase as well as two Shigella spp. (from travellers returning from India); all attempts to transfer resistance by electroporation were unsuccessful and no mutations were detected in their rpoB genes. No arr genes or rpoB mutations were detected in six typhoidal salmonellae and two MDR E. coli isolates with rifamycin MICs of 64 mg/L, suggesting that these represent the high end of the normal wild-type distribution.

4. Discussion Diarrhoea is the most common illness reported in international travellers, particularly those to low-income countries. High levels of resistance are seen to the antibiotics traditionally used to treat travellers’ diarrhoea, principally doxycycline and cotrimoxazole, with increasing resistance to the current first-line agents (azithromycin and ciprofloxacin) among enteropathogens isolated from travellers returning to Spain and the USA [11,12]. Studies to date indicate that rifaximin has good in vitro activity against non-invasive intestinal bacterial pathogens [11–15] but have mainly focused on US travellers visiting Central and South America, destinations associated with a high risk (≥20%) of acquiring travellers’ diarrhoea but accounting for only 1% of visits abroad by UK travellers [16]. Far more UK travellers visit Africa, the Middle East and Asia, all of which are also recognised as high-risk destinations for travellers’ diarrhoea [16]; these accounted for >70% of destinations reported on the referral forms of isolates in this study. There are no clinical breakpoints or published ECOFFs for rifamycins against Enterobacteriaceae but, based on the data presented here, a provisional value of ≤32 mg/L was applied to distinguish wild-type isolates from those likely to have an acquired resistance mechanism. The slightly lower criterion of MIC ≥ 32 mg/L has been used to define resistance among bacterial enteropathogens isolated from travellers’ diarrhoea by Gomi et al. [14] and in studies evaluating the in vitro activity of rifaximin against Clostridium difficile [17–19]. Rifaximin had good in vitro activity against diverse enteropathogenic Enterobacteriaceae isolates; only 8/360 (2.2%) isolates from UK travellers were non-susceptible at >32 mg/L, with no specific resistance mechanism found in the six typhoidal salmonellae with MICs of 64 mg/L, which probably represent the upper end of the normal distribution. All of the 90 E. coli enteropathogenic isolates were judged to be susceptible to rifaximin; this is important because E. coli was identified as the commonest bacterial agent in a systematic review of the global aetiology of travellers’ diarrhoea [20]. These results compared with non-susceptibility rates among enteropathogenic Enterobacteriaceae of 5.6% to azithromycin and 10.0% to ciprofloxacin, both of which are absorbed antibiotics likely to exert wider effects and selection for resistance. Most of the azithromycin resistance in enteropathogenic E. coli was high level (MICs≥128 mg/L). Based

436

K.L. Hopkins et al. / International Journal of Antimicrobial Agents 43 (2014) 431–437

Table 4 Minimum inhibitory concentrations (MICs) (in mg/L) for the recipient strain and two transformants.

Strain Rifampicin Ampicillin Amoxicillin/clavulanic acid a Aztreonam Carbenicillin Cefotaxime/cloxacillin b Cefotaxime Cefotaxime/clavulanic acid a Ceftazidime Ceftazidime/clavulanic acid a Cefpirome Cefpirome/clavulanic acid a Cefoxitin Piperacillin Piperacillin/tazobactam c Temocillin Imipenem/EDTA d Imipenem Meropenem Ertapenem Ciprofloxacin Tobramycin Amikacin Gentamicin Colistin

DH5␣ Select Recipient 8 4 4 ≤0.125 ≤16 ≤0.125 ≤0.125 ≤0.060 0.25 0.25 ≤0.125 ≤0.060 4 ≤1.0 ≤1.0 4 0.25 0.25 ≤0.060 ≤0.125 ≤0.125 0.25 ≤0.5 ≤0.125 ≤0.5

Tr680 Transformant >128 >64 32 ≤0.125 >512 32 128 >32 >256 >32 16 16 >64 32 64 32 0.125 8 1 2 ≤0.125 8 2 1 ≤0.5

Tr338 Transformant >128 >64 >64 8 >512 128 128 >32 >256 >32 64 >32 >64 >64 >64 64 1 16 8 16 0.25 >32 >64 >32 ≤0.5

EDTA, ethylene diamine tetra-acetic acid. Grey shading indicates ≥4-fold increase in MIC compared with the recipient. a Clavulanic acid 4 mg/L, except with amoxicillin where a ratio of 2:1 was used. b Cloxacillin 100 mg/L. c Tazobactam 4 mg/L. d EDTA 320 mg/L.

on these data, rifaximin looks a promising alternative for the types of enteropathogens commonly acquired by UK travellers. However, there must be concern about the potential for rifaximin to cause ‘collateral damage’ by selecting for MDR bacteria of broader public health importance in the gut flora. In this study, 5/90 (5.6%) MDR E. coli isolates with ESBLs and/or carbapenemases exhibited high-level rifaximin resistance (MICs≥256 mg/L), and in two cases this resistance was mediated by arr genes carried on the same plasmids as NDM carbapenemases. We noted a further two MDR E. coli with a His526Asn amino acid substitution in RpoB; this position is within the highly conserved Cluster I of RpoB and is an important contact site for rifampicin [4]. Another MDR E. coli was highly resistant to rifaximin but with no mechanism identified. Rifaximin might therefore select multiresistant strains in the gut flora of travellers, but stronger associations were seen between ESBL and/or carbapenemase production and resistance to ciprofloxacin (61.1%) or azithromycin (44.4%), which implies that these agents are likely to have even greater selectivity for these important multiresistant strains. We found no evidence that rifaximin use would select for MDR S. enterica causing enteric fever. Overexpression of efflux pumps has been shown to contribute to rifaximin resistance in E. coli, with rifaximin MICs of 128 mg/L to ≥256 mg/L [3]. In a recent study by Gomes et al. [21] efflux pumps were the cause of resistance in >95% of rifaximinresistant diarrhoeal and commensal E. coli from children in Peru. Whether efflux contributes to the undefined mechanism(s) of rifamycin resistance observed in the two isolates of Shigella spp. and one MDR E. coli with rifaximin MICs≥128 mg/L requires further investigation. In contrast to Enterobacteriaceae, high rifaximin MICs were noted for most of the Campylobacter spp. screened in this study and have also been observed in Campylobacter spp. isolated from Spanish tourists [11]. Rifamycin resistance is intrinsic to the genus and is mediated by the multidrug efflux pump CmeABC [22,23].

In summary, this study identified lower levels of nonsusceptibility to rifaximin than to alternative antibiotics for travellers’ diarrhoea in enteropathogenic Enterobacteriaceae isolated from UK travellers. Such data suggest that rifaximin may be an option for the treatment of travellers’ diarrhoea caused by noninvasive bacterial pathogens. We also identified a potential concern that rifaximin might select MDR Enterobacteriaceae in the gut flora of travellers; however, this is not an issue unique to rifaximin and stronger associations were seen between ESBL and carbapenemase production and resistance to ciprofloxacin or azithromycin. Moreover, the very high gut concentration of rifaximin may overwhelm resistance, whereas this is unlikely to be the case for absorbed antibiotics.

Funding This work was supported by Norgine Pharmaceuticals Ltd.

Competing interests KLH and SM have received conference support from Norgine Pharmaceuticals Ltd.; KLH, DML and NW have received speakers’ or advisory board honoraria from Norgine Pharmaceuticals; DML consults for numerous pharmaceutical and diagnostic companies, including Meiji Seika, Achaogen, Astellas, AstraZeneca, Bayer, Basilea, bioMérieux, Cubist, Discuva, GSK, Kalidex, Merck, Pfizer, Roche and Tetraphase, holds grants from Basilea, Cubist, Meiji Seika and Merck, has received lecture honoraria or travel reimbursement from AstraZeneca, GSK, J&J, Merck, Novartis, Pfizer and Tetraphase, and holds shares in AstraZeneca, Dechra, Eco Animal Health, GSK, Merck and Pfizer, collectively amounting to

In vitro activity of rifaximin against clinical isolates of Escherichia coli and other enteropathogenic bacteria isolated from travellers returning to the UK.

Rifaximin is licensed in the EU and USA for treating travellers' diarrhoea caused by non-invasive bacteria. Selection for resistance mechanisms of pub...
841KB Sizes 1 Downloads 3 Views