Clinical specimens were processed according to standard microbiological procedures. Specimens were inoculated onto routine culture media, including blood agar and MacConkey agar, and incubated at 35˚C - 37˚C for 18 h - 24 h. Lactose-fermenting Gram-negative bacilli consistent with the order Enterobacteraleswere selected based on colony morphology, Gram stain, and oxidase negativity.

Bacterial identification was performed using VITEK® 2 compact system (bioMérieux, Marcy-L’Etoile, France). A pure colony from a fresh culture was suspended in sterile 0.45% - 0.50% saline, and turbidity was adjusted to a 0.50 McFarland standard using a calibrated densitometer. The Gram-Negative identification card was inoculated according to the manufacturer’s instructions and loaded into the instrument. The VITEK® 2 system incubated and interpreted the card automatically, generating species-level identification based on biochemical reaction profiles. Identification results were accepted when the VITEK® 2 system indicated an acceptable “confidence level”.

Antimicrobial susceptibility testing (AST) was conducted using VITEK® 2 AST cards appropriate for Gram-negative organisms. A standardized suspension (0.50 McFarland) from a pure isolate was prepared in sterile saline and used to inoculate AST cards following the manufacturer’s instructions. The cards were incubated and read automatically by the instrument, which generated minimum inhibitory concentration (MIC) values or categorical interpretations depending on the antimicrobial agent. Susceptibility results were interpreted according to Clinical and Laboratory Standards Institute (CLSI) performance standards current for the year of testing. CLSI categorical breakpoints (susceptible, intermediate, and resistant) were used for all antimicrobial agents studied. Intrinsic resistances and non-reportable agent-organism combinations were identified and excluded from analysis in accordance with CLSI guidelines.

Quality control (QC) procedures were performed following CLSI recommendations using standard reference strains, including Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853. QC was conducted with each new lot of cards and at the frequency of once weekly. AST results were accepted only when QC values fell within CLSI-specified acceptable ranges.

This study delineates the phenotypic characteristics and antimicrobial susceptibility profiles of resistant Enterobacteralesstrains, specifically extended-spectrum β-lactamase-producing Enterobacterales(ESBL-PE) and their associated phenotypes, AmpC-producers, and carbapenem-resistant Enterobacterales(CRE), utilizing the automated VITEK® 2 system (bioMérieux, Marcy-L’Etoile, France). The VITEK® 2 platform has previously been validated as a reliable diagnostic tool for identifying resistant strains with excellent accuracy. For instance, a South African study evaluated the sensitivity and specificity of the VITEK® 2 system for ESBL-PE detection against standard disk diffusion and broth microdilution methods, coupled with conventional polymerase chain reaction (PCR) assays for bla_CTX-M, bla_SHV, and bla_TEM genes. In that evaluation, the VITEK® 2 system demonstrated diagnostic sensitivities for ESBL-PE detection of 92% for Escherichia coli and 100% for Klebsiella pneumoniae, with corresponding specificities of 100% and 90%, respectively. However, that study noted potential diagnostic misinterpretations when evaluating SHV-hyperproducing K. pneumoniae isolates [12].

Concurrently, a multi-center evaluation in North and Latin America compared the VITEK® 2 Advanced Expert System (AES) against genotypic results obtained via whole-genome sequencing (WGS) across 488 resistant Enterobacteralesphenotypes. The VITEK® 2 AES provided phenotypic profiles for 447 (91.6%) isolates, capturing carbapenemases, ESBLs, and plasmid-mediated AmpC (tAmpC) genes, alongside wild-type strains. Overall, the AES phenotypic predictions were correct for 96.9% of the isolates. Specifically, the system accurately identified carbapenemase phenotypes with an overall accuracy of 93.7% (sensitivity: 96.4%; specificity: 91.7%). ESBL-PE phenotypes with an accuracy of 93.7% (sensitivity: 98.1%; specificity: 92.4%). tAmpC phenotypes with an accuracy of 98.4% (sensitivity: 82.1%; specificity: 99.5%). and wild-type isolates with a sensitivity of 100% and specificity of 98.8%. Consequently, these automated phenotypic profiles provide highly reliable therapeutic guidance and significantly reinforce clinical antimicrobial stewardship programs (ASPs) [13].

Only non-duplicate clinical isolates obtained for routine diagnostic purposes were eligible for inclusion. To eliminate bias from repetitive sampling, bacteria included in the final analysis were only the index (first) isolate per a patient, a colonizer, or a pathogen for the first distinct infection episode. Antimicrobial susceptibility testing (AST) data were extracted from the laboratory information system (LIS) and categorized according to the Clinical and Laboratory Standards Institute (CLSI) interpretive breakpoints corresponding to each respective study year. Intermediate isolates were grouped together with resistant strains for analysis. Isolates categorized as susceptible-dose dependent (SDD) were grouped with susceptible strains; however, the total counts for both categories were negligible.

The laboratory receives and process specimens for the inpatients, spanning the different anatomical sources. All specimens are processed for identification and antimicrobial susceptibility testing by VITEK® 2 automated system (bioMérieux, Marcy-L’Etoile, France). The system maintains a multi-year database with a backup repository. Data was exported as text files and formatted into Microsoft Excel sheet (Microsoft Corporation). Patients related metadata collected included hospital location, specimen type, specimen source, specimen date, collection date, testing date, organism name, its bio-number, probability and confidence level. Bacteria isolate-related variables included AST, minimum inhibitory concentrations (MICs), predicted resistance enzyme profiles, and additional Advanced Expert System (AES) algorithmic interpretations of resistance mechanisms across distinct antimicrobials classes.

The Institutional Review Board (IRB) of The Specialty Hospital reviewed and formally approved the study protocol (Approval No. IRB 125771/T/5 issued on 8 June 2026). Because this study did not involve direct human subject participation and relied exclusively on de-identified, historical microbiological data retrieved from the archived VITEK® 2 database, the requirement for informed consent was waived.

The master data frame was converted to a comma-separated values (CSV) format and imported into R version 4.5.2 (R Core Team (2025).R: A language and environment for statistical computing. R Foundation for Statistical Computing,Vienna,Austria. URLhttps://www.R-project.org/.) via the RStudio Integrated Development Environment (Posit team(2025).RStudio: Integrated Development Environment for R. Posit Software,PBC,Boston,MA. URLhttp://www.posit.co/.). Data cleaning, manipulation, and structural adjustments were executed primarily within the “tidyverse” framework. A target subset containing the Enterobacteralesisolates of interest was generated for primary analysis. Descriptive statistics, including frequency distributions, cross-tabulations, and graphical visualizations, were used to characterize baseline resistance patterns, annual isolate distribution rates, and temporal dynamics. To model trends in antimicrobial resistance over the 16-year study period, a generalized linear model (GLM) with a binomial logit link distribution was applied. Logistic regression coefficients were exponentiated to calculate odds ratios (ORs) and their corresponding 95% confidence intervals (CIs). Two-tailed p-values were calculated, and a level of < 0.05 was established as the threshold for statistical significance.

Table 1.Detected resistance mechanisms among common clinical isolates of Enterobacterales.

Note: ^1,*Resistance Phenotype: note that the calculations in the table excluded columns counts with NA values as in E. coli. 1: Total count per tested species. ESBL: Extended Spectrum, B-Lactamases. CRE: Carbapenem-Resistant Enterobacterales. IRT or OXA: Inhibitor-Resistant TEM (IRT) or OXA-type enzymes. PCR or Whole Genome Sequencing was not done to detect the specific resistance gene (blaTEM, blaOXA-1, blaOXA-48, etc.). Molecular method to detect the specific resistance genes (blaKPC, blaNDM, blaIMP, blaVIM, etc.) is needed for a more specific sub type. Impermeabiltiy: impermeable to cabrbapenems. MβL or KPC:Klebsiella pneumoniaemetalo-B-lactamases. TEM or OXA: Inhibitor resistance Temonarian or Klebsiella pneumonia carbapenemases. SHV1_Hyperproduction: Sulphhydril variable causes resistance to penicillins and cephalosporine and affects Ceftazidime (ESBL-like), NA; not tested.

A total of 34,974 bacterial isolates were collected during the 16-year study period. Of these, Gram-negative bacilli accounted for 25,268 isolates, with members of the order Enterobacteralescomprising 20,266 (80.2%) of the Gram-negative cohort.

Among the Enterobacterales{denominator} isolates, Escherichia coli was the most prevalent species (n = 12,783; 63.1%), followed by Klebsiella species—predominantly Klebsiella pneumoniae (n = 4,302; 21.2%), Proteeae species (n = 1,426; 7.0%), Enterobacter species (n = 1,060; 5.2%), Citrobacter species (n = 391; 2.8%), and Serratia species (n = 304; 3.9%). Phenotypic resistance profiles were evaluated both as standalone mechanisms and as co-existing resistance combinations (Table 1).

The overall prevalence of ESBL-PE was 29.8%, primarily driven by Escherichia coli at 47.0% (including 4.7% bla_CTX-M-like variants) and Klebsiella species at 36% (including 2.8% bla_CTX-M-likevariants), with Klebsiella pneumoniae as the dominant species. Minor proportions were identified among Enterobacter species 1.7%, Proteeae 5.3%, Citrobacter species 2.8%, and Serratia species 3.9%, all of which expressed bla_CTX-M-like ESBL phenotypes (Table 1). Longitudinal analysis revealed a stable, minimal downward trend in ESBL-PE prevalence (Figure 1); however, while mathematically significant, the clinical effect size was negligible for both primary organisms (OR = 0.968, SE < 0.006, P < 0.001).

Figure 1.Trends in ESBL and CTX-M expression across Enterobacterales(2010-2025).

Phenotypic AmpC β-lactamase production was 13.9% of the total Enterobacteralescohort. Across individual genera, AmpC production rates were highest in Enterobacter species 25.0%, followed by Klebsiella species 18.0%, Serratia species 19.0%, E. coli 13.0%, Citrobacter species 9.0%, and Proteeae species 6.0% (Table 1).

Chronological trend analysis revealed distinct, divergent paths among specific genera: Klebsiella species demonstrated no trend over time (OR = 1.0, SE = 0.010, P < 0.000), whereas Serratia species displayed a progressive decline (OR = 0.881, SE = 0.037, P = 0.000; Figure 2(a)). E. coli showed no trends (OR = 1.0, SE = 0.00, P < 0.000). Enterobacter (OR = 0.982, SE = 0.017, P = 0.29). Citrobacter (OR = 0.912, SE = 0.048, P = 0.056). Proteeae (OR = 0.906, SE = 0.031, P = 0.002). Cumulatively, the overall trajectory for AmpC-producing strains minimally trending up throughout the study period, showing a significant though a minor and clinically negligible effect size (R = 1.04, SE = 0.006, P < 0.000; Figure 2(b)).

Figure 2. (a) Trends in the prevalence of AmpC-producing Enterobacterales, segregated into species (2010-2025): A single center data; (b) Overall trends for the resistance mechanisms among Enterobacteralesas demonstrated in VITEK-2 (2010-2025): A single-center dataset.

SHV-1 hyperproduction, an advanced β-lactamase expression profile capable of compromising ceftazidime susceptibility was identified in 8.85% of the total cohort. This phenotype was concentrated in E. coli 13.0% and Klebsiella species 1.9% (Table 1). Long-term longitudinal surveillance showed that SHV-1 hyperproduction rates remained highly stable and constant across the multi-year study period (OR = 1.04, SE = 0.006, P < 0.000; Figure 2(b)).

The overall prevalence of the CRE phenotypes was 22.4%. Phenotypic carbapenemase production was most frequent among Klebsiella species 25% and Enterobacter species 11.0%, with lower rates observed in Citrobacter species 4.9%, Serratia species 5.3%, E. coli 3.0%, and Proteeae 2.3% (Table 1).

Longitudinally, CRE rates exhibited a slow, continuous upward trajectory over the 16-year period, though the annual rate of increase was not clinically significant (Figure 3). This steady chronological baseline was consistently reflected across all evaluated taxa: E. coli (OR = 0.998, SE < 0.000, P = 0.0). Klebsiella (OR = 1.00, SE < 0.000, P < 0.000). Enterobacter (OR = 0.999, SE < 0.000, P < 0.000). Proteeae (OR = 0.998, SE < 0.000, P < 0.000). Serratia (OR = 0.999, SE < 0.001, P < 0.000). Citrobacter (OR = 0.999, SE < 0.001, P < 0.001).

Figure 3. Trends in the prevalence of carbapenem-resistant Enterobacterales(CRE; 2010-2025).

To delineate the specific drivers of carbapenem resistance, individual mechanisms were evaluated. Porin impermeability was the most common mechanism, occurring at a total rate of 7.4% across the cohort, driven primarily by Klebsiella 20.3%, Enterobacter 10.1%, and E. coli 2.5%. The cumulative rate for metallo-β-lactamases (MβLs) or Klebsiella pneumoniae carbapenemase (KPC) production was 1.11%, found in 2.4% of Klebsiella, 0.9% of Enterobacter, and < 0.5% in both E. coli and Citrobacter. The inhibitor-resistant TEM or oxacillinase (IRT/OXA) phenotype reached an overall rate of 11.1%, distributed across E. coli 16.0%, Klebsiella 17.0%, and Proteeae 11.0%, while Enterobacter, Serratia and Citrobacter 0.0% (Table 1).

Antimicrobial susceptibility profiles were stratified by specific VITEK® 2-defined phenotypic resistance mechanisms. Isolates sharing identical, discrete resistance mechanisms were grouped together to evaluate susceptibility against antimicrobial agents commonly utilized in clinical practice either as primary definitive therapies or as synergistic components.

For ESBL-PE isolates, overall susceptibility rates ranged from 37.0% to 45.0%. The highest susceptibility rates were observed for aminoglycosides, fluoroquinolones, piperacillin-tazobactam, and tigecycline, whereas the lowest susceptibility rates were noted for aztreonam 37.0% and minocycline 37.0%.

Among AmpC-producing strains, susceptibility rates across all tested antimicrobials were notably lower, ranging from 8.8% to 24.0%. Susceptibility peaked for ceftazidime-avibactam 24.0%, ceftolozane-tazobactam 24.0%, meropenem-vaborbactam 23.0%. Minocycline 13.0%, aztreonam 14.0%. Aminoglycosides, fluoroquinolones, piperacillin-tazobactam, and tigecycline ranged from 8.8% to 12.0%.

For SHV-1 hyperproducers, susceptibility to aminoglycosides, quinolones, and piperacillin -tazobactam ranged from 7.0% to 7.9%. Susceptibility rates for tigecycline, aztreonam, and minocycline clustered between 3.1% and 5.7%.

The collective CRE phenotype demonstrated susceptibility rates for minocycline 34.0%, ceftazidime-avibactam 31.0%, and 32% for ceftolozane/tazobactam, aztreonam was 31.0%, while Aminoglycosides, quinolones, and tigecycline exhibited susceptibilities 12.0% to 18% with tigecycline on the high side. Isolates displaying MβL or KPC phenotypes showed profound multidrug resistance, with susceptibility rates across all tested clinical agents remaining ≤ 0.8%. Strains expressing the TEM or OXA (IRT) phenotype exhibited susceptibility rates ranging from 12.0% to 16.0%, with tigecycline demonstrating the highest activity. Susceptibilities to ceftazidime-avibactam and ceftolozane-tazobactam were remarkably suppressed at 3.5% and 3.6%, respectively.

For isolates whose resistance was mediated by porin impermeability, susceptibility was 30% for aztreonam and minocycline, 23% for meropenem-vaborbactam, Ceftazidime-avibactam 12%, ceftolozane-tazobactam13%, piperacillin-tazobactam 11%, the remaining agents were uniformly suppressed ≤ 10.0%. Finally, evaluating the specific carbapenemase enzyme subgroup revealed susceptibility rates of 90% for meropenem-vaborbactam, 32.0% for both ceftazidime/avibactam and ceftolozane/tazobactam, minocycline 18%, aztreonam 16%. Susceptibility dropped to 12.0% for tigecycline, 11.0% for piperacillin-tazobactam, and 8.5% to 11.0% for aminoglycosides and quinolones (Table 2).

Table 2.Counts and percentages of multidrug-resistant (MDR) Enterobacteralesacross identified resistance mechanisms and their susceptibility to selected antimicrobial agents*.

Note: colistin was not include, it is not approved to be tested by VITEK 2, #: All isolates data, and tested to carbapenem (Imipenem, Ertapenem, Meropenem), CRE: Carbapenem resistant Enterobacterales, 1: n (%), AMK: Amikacin, AZT: Aztreonam, CAZ-AVI: Ceftazidim-Avibactam, CEFT-TAZ: Ceftolozon-Tazobactam, CIP: Ciprofloxacin, GENT: Gentamycin, LEV: Levofloxacin, MER-VAB: Meropenem-Vaborbactam, MIN: Minocycline, PIP-TAZ: Piperacillin-Tazobactam, TIG: Tigecycline, IRT(TEM)orOXA: inhibitor resistance or oxacillinase, MβLorKPC: Metallo-β-lactamase or K.pneumoniae carbapenemase, NA: not tested.