Activity of ceftolozane/tazobactam against surveillance and ‘problem’ Enterobacteriaceae, Pseudomonas aeruginosa and non-fermenters from the British Isles

Abstract Background: We assessed the activity of ceftolozane/tazobactam against consecutive isolates collected in the BSAC Bacteraemia Surveillance from 2011 to 2015 and against ‘problem’ isolates sent to the UK national reference laboratory from July 2015, when routine testing began. Methods: Susceptibility testing was by BSAC agar dilution with resistance mechanisms identified by PCR and interpretive reading. Results: Data were reviewed for 6080 BSAC surveillance isolates and 5473 referred organisms. Ceftolozane/tazobactam had good activity against unselected ESBL producers in the BSAC series, but activity was reduced against ertapenem-resistant ESBL producers, which were numerous among reference submissions. AmpC-derepressed Enterobacter spp. were widely resistant, but Escherichia coli with raised chromosomal AmpC frequently remained susceptible, as did Klebsiella pneumoniae with acquired DHA-1-type AmpC. Carbapenemase-producing Enterobacteriaceae were mostly resistant, except for ceftazidime-susceptible isolates with OXA-48-like enzymes. Ceftolozane/tazobactam was active against 99.8% of the BSAC Pseudomonas aeruginosa isolates; against referred P. aeruginosa it was active against 99.7% with moderately raised efflux, 94.7% with strongly raised efflux and 96.6% with derepressed AmpC. Resistance in P. aeruginosa was largely confined to isolates with metallo-β-lactamases (MBLs) or ESBLs. MICs for referred Burkholderia spp. and Stenotrophomonas maltophilia were 2–4-fold lower than those of ceftazidime. Conclusions: Ceftolozane/tazobactam is active against ESBL-producing Enterobacteriaceae; gains against other problem Enterobacteriaceae groups were limited. Against P. aeruginosa it overcame the two most prevalent mechanisms (up-regulated efflux and derepressed AmpC) and was active against 51.9% of isolates non-susceptible to all other β-lactams, rising to 80.9% if ESBL and MBL producers were excluded.


Introduction
Ceftolozane/tazobactam is a cephalosporin/b-lactamase inhibitor combination, recently licensed for complicated intra-abdominal and urinary tract infections. 1 EUCAST breakpoints, with a fixed 4 mg/L concentration of tazobactam, are susceptible 1 mg/L/resistant .1 mg/L for Enterobacteriaceae and 4/>4 mg/L for Pseudomonas aeruginosa. Tazobactam protects ceftolozane against ESBLs, with the combination's efficacy against producers confirmed in the Phase III trials. 1 Ceftolozane is also notably active against P. aeruginosa, with MICs lower than ceftazidime-hitherto the most active anti-pseudomonal b-lactam. 2 This activity is retained for many strains with derepressed AmpC or up-regulated efflux, 2 which are the major routes to resistance to established penicillins and cephalosporins in the species. 3 Here, we review the activity of ceftolozane/tazobactam against two large, contrasting series of Gram-negative isolates. First, we considered consecutive bloodstream isolates collected by the

Reference submissions
MICs, determined by BSAC agar dilution, 5 were reviewed for all nonfastidious Gram-negative bacteria referred to PHE's AMRHAI Reference Unit over 1 year from July 2015, when routine testing of ceftolozane/tazobactam began. Around 90% of submission are from diagnostic laboratories in England, 9% from elsewhere in the UK and 1% from overseas, principally Ireland. We excluded isolates tested for internal and external quality assurance and repeat/multiple tests on the same isolate from the same submission. Testing employed a wide panel of antibiotics. To predict b-lactamase types, MICs of cefotaxime/clavulanate 2 mg/L, cefotaxime/cloxacillin 100 mg/L, ceftazidime/clavulanate 2 mg/L, ceftazidime/avibactam 2 mg/L, cefepime/clavulanate 2 mg/L and imipenem/EDTA 320 mg/L were compared with those of the same b-lactams alone.
Genes for KPC, VIM, NDM and OXA-48-like carbapenemases were sought by multiplex PCR. Enterobacteriaceae resistant to meropenem and imipenem but lacking genes for these common carbapenemases were subjected to further multiplex PCRs, seeking (i) bla IMP , bla VIM , bla SPM , bla GIM , bla SIM and bla NDM and (ii) bla FRI , bla GES , bla IMI and bla SME . Metallocarbapenemase genes were sought in P. aeruginosa isolates showing 8fold or greater imipenem/EDTA synergy together with broad resistance to penicillins and cephalosporins. bla VEB and bla PER genes were sought by PCR in most P. aeruginosa isolates with ceftazidime MICs .256 mg/L and ceftazidime/clavulanate MICs 32 mg/L. Referred Acinetobacter spp. isolates were screened for bla OXA-51 -like to identify A. baumannii where this gene is universal and chromosomal.

Categorization of isolates
Detection of a b-lactamase gene was considered 'definitive' identification of a mechanism. Carbapenemase-negative isolates were categorized by inhouse mathematical algorithms employing the principles of interpretive reading of phenotypes. 6 Two levels of match were allowed: 'hard', where the phenotype was a perfect match; and 'soft', where the phenotype was less perfectly matched, but the mechanism remained the most likely. For example, to meet hard-match criteria for highly raised (putatively MexAB-OprM-mediated) efflux, a P. aeruginosa isolate required a carbenicillin MIC .512 mg/L with carbencillin:cefotaxime:piperacillin:ceftazidime MIC ratios in the ranges 1:0.25-1:0.03-0.12:0.015-0.06 (note these are ratios, not actual MICs, and are predicated on the fact that up-regulated MexAB-OprM raises MICs of these agents in rough proportion); 4 to meet soft-matchcriteria the strain needed a carbenicillin MIC .512 mg/L and carbenicillin:cefotaxime:piperacillin:ceftazidime MIC ratios in the ranges 1:0.12-2:0.015-0.25:0.008-0.12. Some isolates did not match any of the phenotypes considered and were left as 'unassigned', then categorized according to their level of resistance to reference agents, principally ceftazidime.

BSAC Bacteraemia Surveillance isolates
The BSAC collection provided a random and geographically diverse sample of bloodstream isolates. Because no temporal trend was seen for ceftolozane/tazobactam (not shown), data for 2011 to 2015 were pooled ( Almost all ESBL-and AmpC-producing E. coli (97.9% and 96.6%, respectively) were susceptible to ceftolozane/tazobactam 1!4 mg/L, as were 93.7% of Klebsiella oxytoca hyperproducing K1 b-lactamase. Susceptibility rates for ESBL-producing Klebsiella and Enterobacter were lower, at 84%-85%. Around half the ceftolozane/tazobactam resistance among ESBL-producing Klebsiella was low level, with MICs of 2!4 mg/L; but other isolates were substantially resistant, with MICs up to .256 mg/L. High MICs were not associated with particular ESBL types: 9/13 (69.2%) ESBLproducing isolates with MICs .8 mg/L had CTX-M group 1 enzymes, as did 85/140 (60.7%) with MICs 1 mg/L. It is possible that the more resistant isolates had larger amounts of ESBL, a different CTX-M variant, or multiple b-lactamases. Half of the AmpChyperproducing Enterobacter spp. were resistant; significantly, these ceftolozane/tazobactam-resistant organisms also were much more resistant to cefotaxime and ceftazidime (geometric mean MICs 76.8 and 69.8 mg/L respectively) than were ceftolozane/ tazobactam-susceptible AmpC enterobacters (geometric mean cefotaxime and ceftazidime MICs 6.5 and 5.0 mg/L, respectively). It is likely that this variation reflected the amount of AmpC enzyme.
Activity of ceftolozane/tazobactam JAC One of the two ceftolozane/tazobactam-resistant P. aeruginosa, with an MIC of 32!4 mg/L, had a VIM carbapenemase; mechanisms remain uncertain in the other, with an MIC of 8!4 mg/L.

Isolates referred to PHE
The isolates examined in ceftolozane/tazobactam's first year of testing comprised 3249 Enterobacteriaceae, 1414 P. aeruginosa and 810 other non-fermenters, including 419 Acinetobacter spp. They lack a denominator and are referred for numerous reasons. There is a considerable bias towards submitting isolates suspected of carbapenem resistance, but some have suspected colistin, aminoglycoside or tigecycline resistance; a few are referred owing to anomalous susceptibility, e.g. to ampicillin in P. aeruginosa. The overall performance data shown in Table 2 should be considered with these biases in mind. Nevertheless two points are striking: first, that ceftolozane/tazobactam was more widely active against these 'problem' P. aeruginosa than any other b-lactam, whereas, secondly, gains against problem Enterobacteriaceae were modest, with carbapenems remaining more active. Table 3 shows the MIC distributions of ceftolozane/tazobactam for referred Enterobacteriaceae by species and resistance type. Because there was little difference in their distributions, MICs are pooled for the hard-and soft-matched groups.

Enterobacteriaceae by inferred or proven mechanism
At its 1!4 mg/L breakpoint, ceftolozane/tazobactam was active against 41.7% of the ESBL producers, rising to 53.0% for E. coli and falling to 26.3% for Klebsiella pneumoniae. It was active also against 26.2% of AmpC hyperproducers, rising to 50.8% for E. coli and 88.9% for Morganella morganii, but falling to 18.4% for Enterobacter spp. The high susceptibility rate for AmpCderepressed M. morganii reflects the vulnerability of Morganella AmpC to tazobactam, 7 whilst frequent susceptibility in E. coli was associated with cefotaxime and ceftazidime MICs of 2-4 mg/L, a phenotype usually reflecting small elevations of chromosomal AmpC via promoter mutations. 8 Frequent resistance among the AmpC Enterobacter spp. is in keeping with the fact that 554/649 (85.4%) of these referrals had high-level cefotaxime resistance (MIC 32 mg/L), as typically associated with total derepression of chromosomal enzyme. Klebsiella spp. have no chromosomal ampC, meaning AmpC phenotypes in this genus reflect acquired, Livermore et al.
plasmid-mediated enzymes: a 51.0% susceptibility rate accords with the observation that many examined by PCR (24/26) had bla DHA , a chromosomal 'escape' from M. morganii encoding a tazobactam-inhibited variant. 9 Close relationships existed between ceftolozane/tazobactam MIC and ertapenem MICs for ESBL and AmpC producers, with the percentage of isolates susceptible to ceftolozane/tazobactam falling as ertapenem MIC increased ( Figure 1). Previous experience indicates that raised ertapenem MICs for such carbapenemase-negative isolates mostly reflect porin loss. 10,11 As with the BSAC series, ceftolozane/tazobactam was widely active (6/8 isolates susceptible) against K. oxytoca with high-level K1 enzyme, whereas activity against carbapenemase producers was very limited. All metallo-b-lactamase (MBL)-producing Enterobacteriaceae were resistant, as were almost all with KPC and GES-5 enzymes. On the other hand, 9/12 with rare class A carbapenemases (i.e. IMI, SME or FRI types) remained susceptible, probably because these enzymes are poor cephalosporinases. OXA-48-like carbapenemases are poor cephalosporinases too, especially against ceftazidime. 12 On this basis, we categorized the OXA-48 producers as Ceftolozane/tazobactam was active against around 80% of Enterobacteriaceae isolates inferred to have reduced permeability without ESBL, AmpC or carbapenemase activity-a group also widely susceptible (MIC 1 mg/L, EUCAST) to cefotaxime (59.6%), ceftazidime (48.3%) and cefepime (56.2%). Ceftolozane/ tazobactam also, unsurprisingly, had near-universal activity against cephalosporin-susceptible WTs. For isolates (mostly K. pneumoniae) with unassigned mechanisms, ceftolozane/tazobactam MICs tracked those of unprotected ceftazidime, with 86.8% susceptibility for isolates inhibited by ceftazidime at 4 mg/L, 5.8% for those with ceftazidime MICs of 8-32 mg/L and universal resistance among those with ceftazidime MICs 64 mg/L. It should, however, be stressed that these groups are diverse, and include several phenotypes under active investigation.

P. aeruginosa
The two largest groups of referred P. aeruginosa were those inferred to have increased efflux or derepressed AmpC. The increased efflux group was subcategorized into those with carbenicillin MICs of 256-512 mg/L and those with carbenicillin MICs .512 mg/L ( Table 4). MICs of ceftolozane/tazobactam increased in tandem with those of carbenicillin, piperacillin and ceftazidime for these isolates (Table 4); nevertheless, even among efflux-type isolates with carbenicillin MICs .512 mg/L, 94.7% remained susceptible to ceftolozane/tazobactam 4!4 mg/L, versus 27.6% for ceftazidime and 5.9% for piperacillin/tazobactam. Among effluxtype isolates with carbenicillin MICs 256-512 mg/L, all but one    Hard-matched, isolate's antibiogram conforms precisely to expected phenotype for the mechanism; soft-matched, isolate's antibiogram best matches the phenotype expected for the mechanism, but with minor discrepancies. Livermore et al. Table 4. Hard-matched, isolate's antibiogram conforms precisely to expected phenotype for the mechanism; soft-matched, isolate's antibiogram best matches the phenotype expected for the mechanism, but with minor discrepancies. b bla VEB confirmed by PCR; other ESBL producers were identified by phenotype.
Activity of ceftolozane/tazobactam JAC carbenicillin MICs of 32-128 mg/L and without additional mechanisms affecting piperacillin/tazobactam or ceftazidime were split according to imipenem non-susceptibility (MIC .4 mg/L), which was taken as a marker of OprD loss; ceftolozane/tazobactam MICs were independent of this trait. In contrast to this good activity against isolates with elevated efflux, derepressed AmpC or loss of OprD, resistance to ceftolozane/tazobactam was near universal (96.8%-100%) among P. aeruginosa with MBLs or VEB ESBLs. Isolates with these enzymes also were broadly resistant to other penicillins and cephalosporins, with resistance rates .95%, though 72% of MBL producers remained intermediately susceptible to aztreonam according to EUCAST criteria, with MICs of 2-16 mg/L. Ceftolozane/tazobactam 4!4 mg/L was active against 16/19 GES carbapenemase-positive isolates, compared with 74% for ceftazidime 8 mg/L; however, 14 of these 19 were from a single outbreak, and the results may not be generalizable. Aside from MBL and ESBL producers, the only P. aeruginosa groups where ceftolozane/tazobactam resistance was frequent were the unassigned categories with ceftazidime MICs of 16-128 and 256 mg/L. Ceftolozane/tazobactam MICs for these isolates mostly were 8-16 mg/L, thus slightly lower than for MBL and ESBL producers.
Two wider associations were seen. First, across all 1414 P. aeruginosa, there was broad correlation between the MICs of ceftolozane/tazobactam and ceftazidime, with ceftolozane/tazobactam MICs 2-8-fold lower except for isolates with MBLs or ESBLs. Secondly, (Table 5)  Other non-fermenters MIC distributions of ceftolozane/tazobactam for other nonfermenters besides P. aeruginosa are shown in Table 4. The A. baumannii submissions had a major bias towards carbapenemresistant isolates, with 303/355 non-susceptible to imipenem at its 2 mg/L breakpoint, mostly (.95%) owing to OXA carbapenemases (not shown). Other species tend to be submitted owing to multiresistance, with many of the isolates from cystic fibrosis patients.

Discussion
Ceftolozane/tazobactam combines a new oxyimino-cephalosporin with an established b-lactamase inhibitor. It is licensed for complicated intra-abdominal and urinary tract infections, with efficacy including ESBL-producing Enterobacteriaceae up to the 1!4 mg/L breakpoint. 1,14 Activity against P. aeruginosa is striking, but this species was poorly represented in the licensing trials. We reviewed the combination's activity against: (i) consecutive bloodstream isolates collected in the BSAC Bacteraemia Surveillance; and (ii) 'problem' Gram-negative bacteria submitted to AMRHAI. The latter lack a denominator but provide a snapshot of organisms causing resistance concerns in the UK. For both collections we categorized isolates based on molecular data and interpretive reading.
The BSAC Bacteraemia Surveillance showed ceftolozane/tazobactam to be active against 91.5%-99.7% of the major Enterobacteriaceae species. Activity included 97.9% of ESBL E. coli, but only 84%-85% of ESBL-producing Klebsiella and Enterobacter spp. and 50% of AmpC-derepressed Enterobacter. The latter result is in keeping with tazobactam being a poor inhibitor of Enterobacter AmpC 7 and ceftolozane being a substrate; high resistance rates were likewise seen for referred AmpC-derepressed Enterobacter (Table 3), and have been observed previously. 15 The 15%-16% resistance rates for ESBL-producing Enterobacter and Klebsiella in the BSAC Bacteraemia Surveillance are more surprising, since ESBLs are inhibited by tazobactam. About half this resistance was borderline, with MICs of 2 mg/L, but MICs for the remainder ranged up to 256 mg/L, with resistance not obviously associated with ESBL type. Plausible explanations are that some ESBL-producing Klebsiella and Enterobacter had secondary b-lactamases or permeability lesions. Farrell et al. 16 previously reported 12.1% ceftolozane/tazobactam resistance in 'ESBL phenotype'  (Table 3). They may reflect (i) larger proportions of ESBL producers with secondary mechanisms in the countries surveyed, or (ii) inclusion of strains with KPC enzymes, which can give a false 'ESBL phenotype' in terms of cephalosporin/clavulanate synergy. Frequent resistance to ertapenem among referred ESBL and AmpC producers is probably due to impermeability, and, among these isolates, ceftolozane/tazobactam resistance was common ( Figure 1). Unlike the BSAC collections, the AMRHAI referrals provided numerous carbapenemase-producing Enterobacteriaceae, most of which proved resistant to ceftolozane/tazobactam. This is unsurprising for MBL and KPC producers as (i) MBLs are not inhibited by tazobactam and (ii) KPC enzymes hydrolyse penicillanic acid sulphones, mitigating inhibition. 17 Better activity might be expected against isolates with OXA-48-like enzymes, which are poorly active against oxyimino-cephalosporins, 12 with any cephalosporin resistance arising from secondary mechanisms, principally ESBLs, which should be inhibited by tazobactam. Yet, in reality, ceftolozane/ tazobactam susceptibility was largely restricted to those OXA-48like isolates that were susceptible or intermediate to unprotected ceftazidime (MIC 4 mg/L) and therefore are inferred to lack secondary b-lactamases. Similar behaviour (not shown) was seen for ceftazidime/clavulanate. Plausible explanations, deserving investigation, are: (i) many resistant isolates additionally had permeability lesions or multiple b-lactamases, thus overwhelming the tazobactam; or (ii) OXA-48-like enzymes inactivate tazobactam and clavulanate.
In contrast to this mixed performance against Enterobacteriaceae, ceftolozane/tazobactam was the most active b-lactam against both series of P. aeruginosa isolates. For the BSAC isolates its modal MIC was 4-fold lower than for ceftazidime and the proportion non-susceptible was only 0.2%. More strikingly, ceftolozane/tazobactam was active, at breakpoint, against 99.7% and 94.7% of referred isolates with moderately and strongly increased efflux and against 96.6% with derepressed AmpC. These are the most common resistance mechanisms in P. aeruginosa in the UK 18,19 and Western Europe, [20][21][22] though MBLs and ESBLs, which ceftolozane/tazobactam did not overcome, are more prevalent in Russia, Eastern Europe and the Middle East. [23][24][25] Ceftolozane does not entirely escape efflux and AmpC, and its MICs are slightly raised, nevertheless it is less compromised than other cephalosporins and penicillins, and seems less prone to select for these traits. 26 High-level ceftolozane/tazobactam resistance (MIC .16 mg/L) in P. aeruginosa was largely confined to isolates with MBLs or ESBLs, and these associations were so strong that AMRHAI now uses a ceftolozane/tazobactam MIC .16 mg/L as a predictor of these enzymes for the species. The lack of activity against ESBL-producing P. aeruginosa may seem surprising, since ESBL-producing Enterobacteriaceae were widely susceptible, but we note: (i) the predominant VEB ESBLs may be less susceptible to tazobactam than the CTX-M, TEM and SHV enzymes of Enterobacteriaceae; and (ii) tazobactam may be a poor permeant, or good efflux substrate, for P. aeruginosa. Piperacillin/tazobactam by analogy is poorly active against P. aeruginosa with acquired penicillinases. 3 Ceftolozane/tazobactam offered no gain compared with earlier cephalosporins against Acinetobacter spp. or Elizabethkingia spp., but was more active than ceftazidime against Burkholderia spp. and S. maltophilia. b-Lactams remain controversial in infections due to these species, with medium-dependent MICs 27 and no EUCAST breakpoints. Nevertheless they are often the next-mostactive option after co-trimoxazole in vitro, and may be considered in sulphonamide-intolerant patients or those with co-trimoxazoleresistant strains. Ceftazidime, meropenem and temocillin have the Activity of ceftolozane/tazobactam JAC lowest MICs among b-lactams for Burkholderia spp., 28 with ticarcillin/clavulanate and ceftazidime the most active against S. maltophilia. 29 It now seems appropriate to add ceftolozane/ tazobactam to this list. Activity against S. maltophilia may reflect tazobactam inhibiting the L-2 cephalosporinase, and this possibility deserves exploration.
In countries with conservative antimicrobial usage, such as the UK, new agents enter use as microbiologically directed treatments of problem infections, not as the empirical therapy modelled by Phase III trials. P. aeruginosa is the likely major target for ceftolozane/tazobactam, since b-lactams are core to anti-pseudomonal regimens and the combination inhibited many isolates resistant to all other b-lactams ( Table 5). Results of a high-dosage ventilatorassociated pneumonia trial, where P. aeruginosa is likely to be a frequent pathogen, are awaited with interest. In the meantime there is a growing catalogue of case reports of success in chronic and acute P. aeruginosa infections. [30][31][32][33][34] These data are encouraging, although reports of resistance associated with AmpC sequence variants are a concern. 26,35