Introduction

Extended-spectrum cephalosporin-resistant Escherichia coli (ESCREC) are a serious and growing threat [1], resulting in increasing use of carbapenems for empirical and definitive therapy [2]. Emerging carbapenem resistance in E. coli creates a need to identify suitable non-carbapenem treatment options for ESCREC [3].

Plazomicin (PLZ) is a novel aminoglycoside antibiotic that was approved recently by the US Food and Drug Administration (FDA) for treating complicated urinary tract infection and pyelonephritis [4,5,6,7]. It inhibits protein synthesis by binding to the 30S ribosomal subunit. Its Gram-negative spectrum includes extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae, carbapenem-resistant Enterobacteriaceae (CRE), and organisms with aminoglycoside-modifying enzymes [8,9,10,11]. Thus, PLZ may be a carbapenem-sparing alternative for ESCREC. However, PLZ’s activity specifically against ESCREC has received limited study, especially in relation to clonal background and ESBL genotype.

Antimicrobial resistance in E. coli is highly clonal [12, 13]. The leading ESCREC clonal lineage currently is sequence type ST131 [14, 15], which has multiple distinctive clonal subsets, or subclones [16,17,18]. Of these, H30R, which likewise has multiple important subsets, overall is the most extensively antimicrobial resistant and epidemiologically successful [14].

All H30R members are densely fluoroquinolone resistant, due to four canonical amino acid replacement mutations in gyrA and parC [19]. H30R1 has two main sublineages, H30R1 and H30Rx. H30Rx was recognized first [20], due to its association with the (globally dominant) extended-spectrum beta-lactamase (ESBL)-encoding gene blaCTX-M-15. However, H30R1, the historically “less resistant” sister clade to H30Rx, now has an emerging subclone, C1-M27, that is closely associated with blaCTX-M-27 and blaCTX-M-14 [21]. Unlike blaCTX-M-15 (from blaCTX-M group 1), these two bla variants are from blaCTX-M group 9. Additionally, some non-H30R ST131 strains—especially within the O16-fimH41 subclone (or clade A)—have horizontally acquired ESBL-encoding genes [16,17,18, 22].

Here, we sought to clarify the activity of PLZ against recent ESCREC isolates in comparison with conventional agents, including carbapenems, and in relation to other bacterial characteristics. For that, we determined broth microdilution MICs to PLZ among 216 unique-by-episode ESCREC isolates from the Minneapolis Veterans Affairs Health Care System (MVAHCS) and then compared the MICs statistically with bacterial characteristics, including phylogenetic group, clonal background, blaCTX-M genotype, and susceptibility to 11 relevant comparators.

Methods

Study setting

The MVAHCS is a teaching hospital that provides a full range of patient care services. The MVAHCS clinical microbiology laboratory processes specimens from the Minneapolis campus and 14 outlying community clinics across MN and western WI. Patients are mostly older men, many with multiple chronic medical conditions.

Isolates

From May 2012 through December 2017, with approval from the Institutional Review Board (i.e., Ethics Committee), the research laboratory prospectively collected consecutive E. coli clinical isolates from the MVAHCS clinical microbiology laboratory. In the research laboratory, isolates were stored at −70 C in LB broth supplemented with 20% glycerol.

During the approximately 5.5-year study period, 6324 total E. coli isolates were collected. The clinical laboratory found 267 (4.2%) of these to be resistant or intermediate to ceftazidime and/or ceftriaxone, according to a VITEK-2 instrument (bioMérieux, Durham, NC) and then-current MIC breakpoints, so here were classified as ESCREC. Exclusion of repeat isolates from the same patient within 30 days after an initial isolate left 216 putative unique-by-episode ESCREC isolates as the study population.

By specimen type, the 216 ESCREC study isolates were from (no. of isolates, % of 216) urine (173, 80%); blood (21, 10%); wound (10, 5%); sputum (4, 1.9%); bone (3, 1.4%); fluid (2, 0.9%); and tissue, swab, and other (each: 1, 0.5%). They were derived from 138 unique source patients, with ages ranging from 24 to > 90 years (median, 68 years); 113 (82%) were male.

Susceptibility testing

Isolates underwent standardized broth microdilution MIC determinations with PLZ and, as reported elsewhere [23], 11 comparators, including ertapenem (ETP), imipenem (IPM), and meropenem (MEM), plus eight non-carbapenem agents, i.e., amikacin (AMK), ceftazidime (CAZ), colistin (CLS), gentamicin (GEN), levofloxacin (LVX), minocycline (MIN), tigecycline (TGC), and piperacillin/tazobactam (TZP). Test methods and reference strains were per the Clinical Laboratory Standards Institute (CLSI) [24]. The tazobactam concentration was fixed at 4 mg/L. Interpretive criteria were per CLSI (all agents except TGC and PLZ) or the FDA (TGC and PLZ). Note: the European Committee on Antimicrobial Susceptibility Testing (EUCAST) specifies different MIC breakpoints (mg/L) than does CLSI for GEN (EUCAST, ≥ 4 resistant, vs. CLSI, 8 intermediate) and AMK (EUCAST, ≥ 16 resistant, vs. CLSI, 32 intermediate). Here, isolates with intermediate MIC values were considered resistant.

Molecular typing

As reported elsewhere [23], established PCR-based assays were used to identify E. coli phylogroups A, B1, B2, C, D, E, and F [25]; selected STs associated with multidrug resistance, recent emergence, and/or extraintestinal infections generally [15, 26, 27]; ST131 subsets O16 (clade A), H30R, C1-M27, and H30Rx [27,28,29]; and blaCTX-M. Isolates with blaCTX-M (according to universal blaCTX-M primers [30]) were further characterized with group 1- and group 9-specific multiplex PCR analysis [30].

Fluoroquinolone-resistant ST131-H30 isolates were classified operationally as H30R; H30R isolates that tested negative for H30Rx were classified as H30R1 [27, 28]. All H30R1 isolates were tested for a C1-M27 subclone-specific prophage marker [29] and based on the result were classified operationally as (H30R1) C1-M27 or (non-C1-M27) H30R1.

Statistical methods

Statistical analysis was limited to variables present in ≥ 2 isolates (≥ 1.0 % of 216). Comparisons involving dichotomous variables were tested using chi-squared tests, including an “N-1” chi-squared test for two-group comparisons [31]. Comparisons involving MIC distributions were tested using the Mann-Whitney or Kruskal-Wallis test (two-tailed), due to the nonparametric distribution of the data. Off-scale high or low MICs were analyzed statistically as representing the dilution step above or below (as appropriate) the tested dilution range. For PLZ, MICs were analyzed in relation to susceptibility vs. resistance to each comparator agent that was represented by sufficient resistant (or susceptible) isolates to qualify for statistical analysis.

Summary statistics used for MIC values included the MICmin (lowest detected MIC), MIC50, MIC90, and MICmax (highest detected MIC). Additionally, for PLZ, GEN, and AMK, cumulative percent MIC distributions were tabulated. Throughout, the criterion for statistical significance was P < 0.05, without adjustment for multiple comparisons, given the study’s exploratory nature. For MIC comparisons that yielded a statistically significant difference despite similar or identical MIC summary statistics, mean MIC ranks (note: not mean MICs) were used to clarify the direction of the difference.

Results

Overall susceptibility

Of the 12 study agents, four (PLZ, IMP, MEM, and TGC) exhibited > 99% susceptibility (Table 1). These were followed, in descending order, by AMK (98%), ETP (96%), MIN (75%), GEN (64%), CAZ (36%), and LVX (12%). No isolate—by definition (per CLSI)—was susceptible to CL (Table 1). PLZ exhibited the lowest MIC50/MIC90 ratio (i.e., 4) of all study agents excepting LVX (MIC50 and MIC90 both > 8 mg/L; ratio uninterpretable).

Table 1 Overall percent susceptible and MICmin, MIC50, MIC90, and MICmax for plazomicin and 11 comparators among 216 extended-spectrum cephalosporin-resistant Escherichia coli clinical isolates
Full size table

Phylogroups

PLZ MICs did not vary significantly across phylogroups, by contrast with MICs for all but three comparators (IPM, GEN, AMK) Suppl. Table (1). The PLZ MIC50 was 1 mg/L for all phylogroups except phylogroup C (0.5 mg/L), and the PLZ MIC90 was consistently 2 mg/L. The cumulative percent MIC distribution for PLZ, GEN, and AMK likewise showed minor variation across phylogroup Suppl. Table (2).

Table 2 Distribution of MIC50 and MIC90 by sequence type and ST131 subclone for plazomicin and 11 comparators among 216 extended-spectrum cephalosporin-resistant Escherichia coli clinical isolates
Full size table

STs

PLZ MICs also did not vary significantly across the four most prevalent STs and all other STs combined, by contrast with MICs all but four of the comparators (IPM, GEN, AMK, TGC) (Table 2). Within each ST category, the PLZ MIC50 was consistently 1 mg/L and the PLZ MIC90 consistently 2 mg/L. The cumulative percent MIC distribution for PLZ, GEN, and AMK showed minor variation across STs (Supplemental Table 3).

ST131 subclone

By contrast with phylogroup and ST, PLZ MICs did vary significantly by ST131 subclone status, albeit subtly, as reflected in mean MIC ranks (Table 2 footnote) and the cumulative percent MIC distribution (Supplemental Table 4), not the MIC50 or MIC90 (Table 2). Specifically, according to mean MIC ranks, PLZ MICs were lowest among the O16 (clade A) ST131 isolates; highest among the (H30R1) C1-M27 ST131 isolates; and intermediate among the non-ST131 isolates, the non-C1-M27 H30R1 isolates, and the H30Rx isolates. Likewise, MICs for all but three comparators (MEM, AMK, TGC) also varied significantly by ST131 subclone status, in agent-specific patterns.

CTX-M genotype

PLZ MICs did not vary significantly by CTX-M status or CTX-M group (Supplemental Table 5). By contrast, MICs did so vary for eight of the 11 comparators (i.e., all but MEM, GEN, and TGC). For these eight agents, MICs were usually higher among CTX-M-positive isolates, and/or CTX-M group 1-positive isolates. The cumulative percent MIC distribution for PLZ, GEN, and AMK illustrated with greater granularity this distinction between GEN (higher MICs among CTX-M and group 1-positive isolates) vs. PLZ and AMK (minimal variation in relation to resistance genotype) (Supplemental Table 6).

Resistance to comparators

PLZ MICs varied significantly with comparator-agent-resistance only for AMK or LVX, in both instances being higher among comparator-resistant isolates (Supplemental Table 7). With AMK, the fact that PLZ MICs were higher among resistant isolates was reflected in the twofold higher MIC50 (2 mg/L, vs. 1 mg/L). By contrast, with LVX, the direction of the difference was evident only from mean MIC ranks (88, LVX-susceptible isolates; vs. 111, LVX-resistant isolates: Supplemental Table 7 footnote).

Discussion

In this study of the activity PLZ and 11 comparators against 216 ESCREC clinical isolates from veterans in relation to bacterial characteristics, PLZ distinguished itself from most or all comparators in multiple respects. These included (i) a very high overall percent susceptible (99.5%); (ii) very similar MIC50 and MIC90 values (only a four-fold difference); (iii) minimal MIC variation in relation to phylogroup, ST, or beta-lactamase genotype; and (iv) subtle but statistically significant MIC variation in relation to ST131 subclone status (lowest among O16 isolates, highest among C1-M27 H30R1 isolates). PLZ MICs were also independent of resistance to comparator agents, except for AMK (possibly due to shared resistance mechanisms [32]) and LVX (possibly due to clonally or genetically linked resistance mechanisms [33]). These findings demonstrate the distinctiveness of PLZ and its preserved activity against even multidrug-resistant ESCREC isolates, and recommend it as a potential non-beta-lactam, carbapenem-sparing alternative for treating ESCREC infections.

The high percent susceptible for most agents obliged comparisons based on MICs rather than percent susceptible. In these analyses, multiple comparator agents—but rarely PLZ—exhibited MIC shifts in relation to each category of variable studied. The observed MIC differences within the susceptible range conceivably could be clinically significant, depending on the site and severity of infection [34,35,36,37], if drug levels at the site were limited by local or systemic factors, or with immune compromise.

The only phylogenetic entity to exhibit comparatively higher PLZ MICs was the recently recognized and emerging C1-M27 subset within ST131-H30R1, which is associated with blaCTX-M-27 and blaCTX-M-14 [21, 29, 38]. The basis for the higher PLZ MICs of C1-M27 isolates is unclear. Conceivably, the same plasmids that carry blaCTX-M-27 may carry genes that encode resistance mechanisms (e.g., ribosomal methyl-transferases or efflux pumps) that raise PLZ MICs, without conferring full resistance.

Study limitations include the single-institution source of the isolates (MVAHCS); the distinctiveness of the veteran population, which may constrain generalizability; the minimal data regarding the source patients and their clinical presentations; the lack of information regarding PLZ resistance mechanisms; and the uncertain therapeutic implications of the MIC data. Study strengths include the relatively large and recent sample, the extensive molecular and phenotypic characterization of the isolates, and the analysis of MICs in relation to multiple bacterial characteristics, including resistance to comparators.

In conclusion, we found that PLZ exhibited activity against recent ESCREC clinical isolates comparable to that of carbapenems and that most bacterial characteristics were unassociated with shifts in the PLZ MIC, by contrast with the extensive associations of these variables with MICs for most comparators. These findings support PLZ as a potential alternative to carbapenems for treating ESCREC infections, largely irrespective of phylogenetic/clonal background or ESBL genotype, and support further attention to PLZ susceptibility within the emerging C1-M27 ST131 subclone.