Activity of ceftolozane-tazobactam against Escherichia coli isolates from U.S. veterans (2011) in relation to co-resistance and sequence type 131 (ST131) H30 and H30Rx status

Background Escherichia coli sequence type 131 (ST131), with its multidrug-resistance-associated H30R1 and H30Rx clonal subsets, causes most antimicrobial-resistant E. coli infections in the U.S., especially among veterans. The activity of ceftolozane-tazobactam (C/T), a new beta-lactamase inhibitor agent, against ST131 strains, and E. coli isolates from veterans, is undefined. Methods We determined broth microdilution MICs for C/T and five comparators–piperacillin-tazobactam (TZP) levofloxacin (LVX), gentamicin (GEN), ceftazidime (CAZ), and meropenem (MEM)–for 595 clinical E. coli isolates, collected in 2011 from 24 Veterans Affairs Medical Centers across the U.S. Categorical resistance and MICs were compared statistically with resistance category (fluoroquinolone-susceptible, fluoroquinolone-resistant, and extended-spectrum beta-lactamase [ESBL]-producing) and with PCR-defined ST131, H30R1, and H30Rx status. Results Resistance prevalence was ≤ 6% for C/T (6%) and MEM (0%), vs. from 8.0% (TZP) to 59% (LVX) for the other comparators. MICs generally increased by resistance category, from fluoroquinolone-susceptible through fluoroquinolone-resistant to ESBL, and by clonal subgroup, from non-ST131-H30 through H30R1 to H30Rx. For each comparator agent except MEM, although a significantly greater fraction of resistant than susceptible isolates were C/T-resistant, only a minority of comparator-resistant isolates were C/T-resistant (i.e., 9% if LEV-resistant, 12% if GEN-resistant, 21% if CAZ-resistant, 38% if TZP-resistant). Conclusions C/T was broadly active against E. coli clinical isolates from veterans, notwithstanding significant variation by resistance category and ST131-H30R1/H30Rx status, outperforming all non-carbapenem comparators. C/T should prove useful as a carbapenem-sparing therapy for multidrug-resistant E. coli ST131 infections, including in veterans.

The combination agent ceftolozane/tazobactam (C/T) combines ceftolozane, a new antipseudomonal cephalosporin, with tazobactam, a beta-lactamase inhibitor that extends ceftolozane's spectrum of activity to include ESBL-producing and other multidrug-resistant (MDR) pathogens [11]. Based on the positive results of phase 3 clinical trials, C/T received FDA approval for treatment of complicated urinary tract infections and complicated intra-abdominal infections [12].
In recent studies C/T was active against most E. coli clinical isolates, although these derived primarily from non-U.S. locales [13][14][15]. The activity of C/T specifically against ST131 strains, especially those from U.S. veterans, is unknown. Given the importance of ST131-H30R1 as a driver of the current MDR E. coli pandemic [16], the tendency with some antimicrobial agents for resistant ST131 strains to have higher MICs and MBCs than other resistant E. coli [10], and the prominence of ST131 among veterans [1], we sought to determine C/T susceptibility in relation to both co-resistance phenotypes and ST131 subclone status among E. coli isolates collected from 24 Veterans Affairs medical centers (VAMCs) across the U.S. in 2011.

Isolates
As reported elsewhere [1], representative sets of E. coli clinical isolates were collected in 2011 from 24 VA medical centers distributed widely across the U.S. Each center provided~10 each consecutive fluoroquinolone-susceptible (FQ-S) and fluoroquinolone-resistant (FQ-R) isolates (~20 total), without regard for other resistance phenotypes (including possible cephalosporin resistance or ESBL production). Additionally, due to the comparative rarity of ESBL-producing isolates, each center also provided up to 10 such isolates from 2010-2011. These isolates were classified as ESBL-producing by the source laboratories based on each laboratory's thencurrent methods and criteria. In total, 595 isolates were collected (within three resistance categories: 234 FQ-S, 238 FQ-R, and 123 ESBL).

Susceptibility testing
We newly determined broth microdilution MICs for C/T and five comparator agents, including piperacillin-tazobactam (TZP), levofloxacin (LVX), gentamicin (GEN), ceftazidime (CAZ), and meropenem (MEM), by using Clinical and Laboratory Standards Institute (CLSI) specified procedures and reference strains [18]. CLSI-specified breakpoints were used to assign categorical interpretations with intermediate and resistant considered as resistant [19]. For C/ T and CAZ the initial concentrations tested were from 0.25 to 32 mg/L, in doubling dilutions. If growth was seen at 32 mg/L, additional doubling dilutions were tested, up to 256 mg/L. The dilution range for other agents as follows: TZP (2.0 to 256 mg/L), GEN (0.25 to 32 mg/L), LEV (0.125 to 16 mg/L) and MEM (0.016 to 8 mg/L).

Statistical methods
Comparisons were tested using a chi-square test for dichotomous variables and the Mann-Whitney U test for continuous variables. P < .05 was the criterion for statistical significance.

Categorical resistance, overall and by clonal subset
Among the 595 total study isolates, resistance prevalence values ranged by agent from 0% (MEM) to 59% (LVX); C/T exhibited the second lowest value (6%) ( Table 1). For each agent except MEM, resistance prevalence tended to increase by resistance category, from FQ-S through FQ-R to ESBL, and by clonal subgroup, from non-H30R1 through H30R1 to H30Rx. The two exceptions to this trend that involved absolute prevalence differences of > 5% included a slightly higher prevalence (i) of LVX resistance among the FQ-R isolates (100%, by definition) than among the ESBL isolates (92%), and (ii) of GEN resistance among the H30R1 isolates (42%) than among the H30Rx isolates (32%). Nearly all resistance prevalence differences across resistance categories and clonal subgroups were statistically significant (Table 1).

MICs in relation to resistance category
Overall, MICs ranged from below to above the boundaries of the tested drug concentration range for each agent except GEN (minimum, 0.5 mg/L) and MEM (maximum, 1.0 mg/L) ( Table 2) (Table 2).

MICs in relation to clonal subgroup
The three clonal subgroups (non-ST131-H30, H30R1, and H30Rx) exhibited significant MICs differences for all studied agents except MEM, including in 5 of 6 three-group comparisons and in 12 of 18 pairwise comparisons ( Table 3). The rank order of subgroups for MICs varied by agent. For C/T, exceptionally, the rank order of the subgroups (by increasing MICs) was

Prevalence of C/T resistance vs. resistance to comparator agents
The prevalence of co-resistance to C/T among isolates resistant to TZP, CAZ, GEN, or LEV ranged by comparator agent from 9% (LEV) to 38% (TZP), and was significantly higher among such isolates than among isolates susceptible to the particular agent (Table 4). C/T coresistance was more frequent among isolates resistant to other beta-lactam agents (TZP, 38%; CAZ, 21%) than among those resistant to non-beta-lactam agents (GEN, 12%; LEV, 9%).

Discussion
Here, among 595 recent E. coli clinical isolates from U.S. veterans (2011) [1], we assessed MICs and categorical resistance for the novel combination agent C/T and 5 comparator agents, then compared these results statistically with resistance category (FQ-S, FQ-R, and ESBL) and ST131-based clonal subgroup (non-H30, H30R1, and H30Rx). Our findings support four main conclusions. First, C/T was broadly active, encountering resistance in only 6% of study isolates overall (despite deliberate enrichment for resistant strains), which was the lowest value among the five non-carbapenem study agents. Second, C/T MICs and categorical resistance prevalence increased significantly by resistance category, from FQ-S through FQ-R to ESBL. Third, for each comparator agent the great majority of resistant isolates remained susceptible to C/T. Fourth, although susceptibility to C/T and the comparators varied significantly in relation to clonal background, with H30Rx isolates being most resistant, C/T performed well even against H30Rx isolates (16% C/T-resistant; C/T MIC 90 = 8.0). Regarding the observed 6% overall resistance prevalence for C/T, even this favorable statistic-which would qualify C/T for use in empirical treatment of pyelonephritis according to Infectious Diseases Society of America guidelines [20]-doubtless overstates the actual C/T resistance prevalence among unselected clinical E. coli isolates from veterans. This is because the study population was deliberately enriched for FQ-R and ESBL isolates, which our results show are more likely to be C/T-resistant than are other E. coli.
Regarding the progressive increase in categorical resistance to C/T and C/T MICs across the three resistance categories (FQ-S, FQ-R, and ESBL), this is consistent with the known presence within the FQ-R group of an ESBL-producing subset, with some ESBLs, possibly in conjunction with other resistance mechanisms, likely conferring resistance to C/T. Despite these trends, C/T still almost always exhibited better susceptibility statistics than did the other noncarbapenem study agents, including within the ESBL isolate group (21% C/T resistance, vs. 18%-92% for the non-MEM comparator agents). The highest observed resistance prevalence for C/T, which was among TZP-resistant isolates, was only 38%, indicating that C/T could serve as an alternate agent for nearly two-thirds of TZP-resistant isolates, let alone for nearly all isolates resistant to the other non-carbapenem comparators.
Regarding the clonal distribution of resistance to C/T and the comparators, the observed patterns were consistent with class-specific resistance mechanisms that are known to be clonally associated [5,17,21,22]. Specifically, for beta-lactams (C/T, TZP, and CAZ), resistance prevalence and MICs were higher among (ESBL-associated) H30Rx isolates than among the non-H30 and the H30R1 isolates, which did not differ greatly from one another. By contrast, for GEN and LEV, resistance prevalence and MICs were higher among the H30R1 and the H30Rx isolates (which did not differ greatly from one another) than among the non-H30 isolates.
An additional unexpected finding was the statistically significantly (albeit subtly) higher C/ T MICs among the non-H30 isolates than the H30R1 isolates, as reflected in the upper limit of the range (Table 3). This may have been due to a larger ESBL-producing subset among the non-H30 isolates than among the H30R1 isolates, although such a difference is not apparent from the susceptibility data for TZP or CAZ. Different results could be anticipated in regions where a substantial fraction of H30R1 isolates produce CTX-M-14 or CTX-M-27 [6,23].
The study is limited by its in vitro nature, since MICs may not correspond with clinical efficacy, its focus on veterans (who may not represent well the general U.S. population), the lack of attention to genetic resistance mechanisms, the uncertain current relevance of isolates from 2011, and the geographic restriction to the U.S. Study strengths include the broadly distributed (within the U.S.) strain collection, the comparisons with the clinically relevant ST131 clonal subgroups H30R1 and H30Rx, and the inclusion (for both MICs and categorical resistance) of a range of relevant comparator agents.
In summary, within broadly representative set of clinical E. coli isolates from U.S. veterans from 2011 we found that C/T is broadly active, albeit with significant variation in relation to co-resistance and to H30R1 and H30Rx subclone status, and outperformed all non-carabapenem comparators. Thus, C/T should prove useful as a carbapenem-sparing agent against multidrug-resistant E. coli infections in veterans, including those caused by ST131-H30R1 and H30Rx strains.