Reconciling the Potentially Irreconcilable? Genotypic and Phenotypic Amoxicillin-Clavulanate Resistance in Escherichia coli

Abstract

Resistance to amoxicillin-clavulanate, a widely used beta-lactam/beta-lactamase inhibitor combination antibiotic, is rising globally, and yet susceptibility testing remains challenging. To test whether whole-genome sequencing (WGS) could provide a more reliable assessment of susceptibility than traditional methods, we predicted resistance from WGS for 976 Escherichia coli bloodstream infection isolates from Oxfordshire, United Kingdom, comparing against phenotypes from the BD Phoenix (calibrated against EUCAST guidelines). A total of 339/976 (35%) isolates were amoxicillin-clavulanate resistant. Predictions based solely on beta-lactamase presence/absence performed poorly (sensitivity, 23% [78/339]) but improved when genetic features associated with penicillinase hyperproduction (e.g., promoter mutations and copy number estimates) were considered (sensitivity, 82% [277/339]; P < 0.0001). Most discrepancies occurred in isolates with MICs within ±1 doubling dilution of the breakpoint. We investigated two potential causes: the phenotypic reference and the binary resistant/susceptible classification. We performed reference standard, replicated phenotyping in a random stratified subsample of 261/976 (27%) isolates using agar dilution, following both EUCAST and CLSI guidelines, which use different clavulanate concentrations. As well as disagreeing with each other, neither agar dilution phenotype aligned perfectly with genetic features. A random-effects model investigating associations between genetic features and MICs showed that some genetic features had small, variable and additive effects, resulting in variable resistance classification. Using model fixed-effects to predict MICs for the non-agar dilution isolates, predicted MICs were in essential agreement (±1 doubling dilution) with observed (BD Phoenix) MICs for 691/715 (97%) isolates. This suggests amoxicillin-clavulanate resistance in E. coli is quantitative, rather than qualitative, explaining the poorly reproducible binary (resistant/susceptible) phenotypes and suboptimal concordance between different phenotypic methods and with WGS-based predictions.

Keywords: antibiotic resistance; antimicrobial combinations; beta-lactamase inhibitor; microbial genomics; susceptibility testing.

FIG 1

Association between transmissible beta-lactamase gene DNA copy number and amoxicillin-clavulanate MIC in isolates with no alternative resistance features. Evidence for the association between MIC and log₂(DNA copy number) (P < 0.0001, estimated using quantile regression), is presented. The gray line indicates the 2.5 threshold used to define resistance in the extended algorithm based on ROC analysis (see Fig. S3A in the supplemental material). Of these 328 isolates, 294 had bla_TEM genes (290 with bla_TEM-1, 4 with other non-inhibitor resistant bla_TEM genes), 19 had non-inhibitor-resistant bla_SHV genes, and 15 had bla_CTX-M genes.

FIG 2

Proportion WGS predicted resistant (extended algorithm) by routine laboratory MIC.

FIG 3

Comparison of the three different phenotyping methods on the agar dilution subsample isolates (n = 261). MICs obtained using three different phenotyping methods were compared: EUCAST-based agar dilution, CLSI-based agar dilution, and BD Phoenix (performed in the OUH microbiology laboratory and using panels calibrated against EUCAST guidelines). *, isolates are in categorical agreement if they are reported as either resistant by both methods (i.e., BD Phoenix/EUCAST-based agar dilution MIC > 8/2 mg/liter and CLSI-based agar dilution > 16/8 mg/liter) or susceptible by both methods (i.e., BD Phoenix/EUCAST-based agar dilution MIC ≤ 8/2 mg/liter and CLSI-based agar dilution MIC ≤ 8/4 mg/liter). Intermediate isolates were excluded from these comparisons (but are shown above) since BD Phoenix/EUCAST-based agar dilution has no intermediate category. Blue, full agreement of MICs; light orange, essential agreement; dark orange, within two doubling dilutions (theoretically feasible believing both tests having an error of ± 1 dilution); red, disagreement.

FIG 4

Proportion WGS predicted resistant (extended algorithm) by MICs from EUCAST and CLSI-based methods. In the main panel, each (x and y) coordinate represents (EUCAST-based MIC, CLSI-based MIC) combination. At each coordinate, the circle size represents the number of isolates with this combination of fixed and ratio MICs, and the color denotes the proportion identified as resistant by WGS, as indicated by the color bar to the right of the figure. The two subpanels (bar charts to the left and bottom of the main panel) show the number of isolates with each MIC (in line with the main panel). Yellow/blue coloring indicate which of these were predicted resistant/susceptible respectively, and black lines indicate cutoffs used to determine resistance classification (susceptible/resistant for EUCAST-based agar dilution, susceptible/intermediate/resistant for CLSI-based agar dilution).

FIG 5

Changes in doubling dilution MIC independently associated with each feature/testing method (multivariable random-effects model). Purple represents testing using 2:1 CLSI-based agar dilution (CLSI), and green represents testing using EUCAST-based agar dilution. All elements except those denoted by an asterisk (*) and shaded in orange are modeled as binary presence versus absence effects (see the supplementary methods): other_bla (grouped other bla genes, includes bla_TEM-40 [n = 2], bla_TEM-30 [n = 3], bla_CMY-2 [n = 3], bla_OXA-48 [n = 1], bla_TEM-190 [n = 1], bla_TEM-33 [n = 1]; Table S2 in the supplemental material), blaOXA:2d (Bush-Jacoby 2d, bla_OXA), blaCTXM:2be (Bush-Jacoby 2be, CTXM), blaTEM:2b (Bush-Jacoby 2b, bla_TEM), blaSHV:2b, (Bush-Jacoby 2b, SHV), ampCpr (ampC promoter mutation suggesting increased expression), blaTEMpr (bla_TEM hyperproducing promoter), NFOMP (nonfunctional ompF/ompC), blaTEM:2b:cpn (copy number) effect modeled as effect of doubling copy number, and blaSHV:2b:cpn (copy number) effect modeled as effect of doubling copy number.

FIG 6

Model-based MIC prediction for non-subsample isolates (n = 704). Blue shading indicates correctly predicted observed AST MIC (554/704 [79%] isolates), light pink indicates predicted within one doubling dilution (total 683/704 [97%] isolates, essential agreement), orange indicates within two doubling dilutions (total 701/704 [100%]), and red indicates >2 doubling dilutions. The results shown exclude eleven isolates with resistance mechanisms not included in the agar-dilution subsample on which the prediction model was derived (similar overall performance including these).