Phenotypic and Molecular Characterization of ESBL/pAmpC-Associated Resistance in Poultry- and Hatchery-Derived Escherichia coli in Bosnia and Herzegovina

Abstract

Antimicrobial resistance (AMR) in poultry-associated Escherichia coli (E. coli) is a persistent One Health concern, particularly when ESBL/pAmpC determinants co-occur with resistance to multiple antimicrobial classes. Between March and October 2024, we investigated commensal E. coli from three interconnected compartments of the poultry production chain in Bosnia and Herzegovina (parent-breeder flocks, commercial broiler farms, hatchery-associated material). A total of 333 samples were examined, and 99 E. coli isolates were recovered (29.7%). Phenotypic characterization included ESBL confirmation, disk diffusion susceptibility testing, and EUVSEC broth microdilution. Targeted real-time PCR assays were used to screen key ESBL/pAmpC-associated genes and selected carbapenemase and plasmid-mediated colistin resistance targets within the targeted panel. ESBL phenotypes were detected in 52/99 isolates (52.5%), and multidrug resistance was highly prevalent across compartments (93/99; 93.9%). ESBL/pAmpC-associated genes were detected in 91/99 isolates (91.9%), with bla_TEM predominating. Gene pattern analysis indicated that bla_TEM occurred most frequently as a single determinant and as part of the predominant multi-gene combinations, most notably bla_TEM + bla_CMY and bla_TEM + bla_CTX-M, while bla_SHV was sporadic. Carbapenemase genes (bla_KPC, bla_NDM, bla_GES, bla_OXA-48) and mcr-1 to mcr-9 were not detected. Overall, our findings indicate a substantial ESBL/MDR burden throughout the poultry production chain, supporting the need for strengthening antimicrobial stewardship and biosecurity measures across both farms and hatcheries.

1. Introduction

Poultry production remains one of the most important segments of animal agriculture worldwide and is a major source of animal protein for human consumption [1]. In parallel with sector growth, management practices are evolving, including increased interest in antibiotic-free or reduced-antibiotic production systems in some regions. These shifts support antimicrobial stewardship goals, but they also place greater emphasis on prevention and control of bacterial diseases at the flock level [2].

Among bacterial pathogens affecting poultry, Escherichia coli (E. coli) is particularly relevant because it is ubiquitous and highly diverse, and it can act both as a commensal organism and as an extraintestinal pathogen [3]. Avian pathogenic E. coli (APEC) is a key etiological agent of colibacillosis and is associated with respiratory and systemic disease manifestations, welfare impairment, and economic losses in commercial flocks [3,4]. In addition, the literature increasingly discusses the broader One Health relevance of poultry-associated E. coli lineages, particularly in the context of antimicrobial resistance and potential exposure pathways linked to poultry production and poultry products [3,5]. While APEC is a major clinical driver of colibacillosis, commensal/indicator E. coli is widely used as a practical proxy for monitoring the antimicrobial resistance reservoir circulating within the poultry production chain and its potential One Health relevance [3,5,6,7].

The structure of the poultry production chain is important for understanding E. coli epidemiology. Evidence supports that dissemination may occur across interconnected production stages, including breeder flocks, hatcheries, and broilers, through combinations of vertical and horizontal transmission routes [8,9]. Hatcheries represent a critical node, as early-life exposure can facilitate rapid spread within batches and between production cycles, which may influence downstream health outcomes in broilers [8,9].

Antimicrobial resistance (AMR) in poultry-associated E. coli has become a major concern, particularly due to plasmid-mediated mechanisms that support persistence and dissemination [5,6]. Of specific importance are extended-spectrum β-lactamase (ESBL) and plasmid-mediated AmpC (pAmpC) phenotypes, which compromise clinically important β-lactams and can disseminate along the production chain [6,7]. β-Lactam antimicrobial resistance in Enterobacterales is predominantly mediated by β-lactamases, commonly classified by the Ambler molecular scheme into classes A, B, C, and D (serine β-lactamases in classes A/C/D and metallo-β-lactamases in class B). ESBLs are most often class A enzymes (e.g., TEM-, SHV-, and CTX-M-type variants) that hydrolyze extended-spectrum (oxyimino-)cephalosporins and aztreonam and are generally inhibited by clavulanic acid. In contrast, AmpC β-lactamases (class C), including plasmid-mediated AmpC, additionally compromise cephamycins (e.g., cefoxitin) and are typically poorly inhibited by clavulanate [10,11]. European monitoring and reviews consistently report ESBL/pAmpC-producing E. coli in broiler production, with substantial variation between countries and sampling points. ESBL enzymes typically confer resistance to penicillins, extended-spectrum (oxyimino-)cephalosporins (e.g., third-generation cephalosporins), and monobactams (aztreonam), and are generally inhibited by clavulanic acid. In contrast, AmpC enzymes also compromise cephamycins (e.g., cefoxitin) and are poorly inhibited by clavulanate, while carbapenems usually remain stable to both enzyme classes [6,7].

In Bosnia and Herzegovina, published data confirm the presence of ESBL/pAmpC-associated E. coli in poultry-related settings, yet production-stage-specific evidence remains limited [12,13]. Therefore, this study investigated E. coli from key points of the poultry production chain in Bosnia and Herzegovina (parent-breeder flocks, broiler flocks, and hatchery-associated material). Specifically, we assessed phenotypic antimicrobial susceptibility (including ESBL confirmation and multidrug resistance patterns) and performed targeted molecular screening of selected resistance determinants relevant to ESBL/pAmpC and plasmid-mediated colistin resistance, aiming to evaluate phenotype–genotype concordance and provide evidence that supports control measures for an avian pathogen of major animal health importance and potential One Health relevance.

2. Materials and Methods

2.1. Study Populations and Sample Collection

Sampling was conducted between March and October 2024 across three poultry production compartments in Bosnia and Herzegovina: a parent-breeder farm (heavy line), commercial broiler farms, and a hatchery. Parent-breeder sampling was performed on a COBB 500 heavy-line farm consisting of six buildings (10,000 birds per building; 60,000 birds per production cycle). The broiler component comprised 163 farms, with sampling performed when broilers were 27–35 days old. Boot swab sampling was performed using two pairs of boot swabs for broiler flocks and five pairs for parent-breeder flocks. Boot swabs with adequate absorbency were pre-moistened with 0.8% saline solution or sterile water, and material was collected by walking through all areas of the house, including littered zones, to obtain representative samples. Hatchery samples consisted of yolk sacs and parenchymatous organs collected aseptically from dead and culled day-old chicks in the hatchery. All samples were transported to the Laboratory for molecular–genetic and forensic investigation, Veterinary Institute, University of Sarajevo–Veterinary Faculty, in sealed containers while maintaining the cold chain and were processed immediately upon arrival.

2.2. Isolation and Identification of E. coli

Isolation was performed using a standard bacteriological workflow as described by Quinn et al. (2011) [14]. Briefly, pooled boot swabs with feces and hatchery organ material were pre-enriched in buffered peptone water (BPW) (Condalab, Madrid, Spain) at a 1:9 ratio and incubated aerobically at 37 °C for 18 ± 2 h. After incubation, cultures were streaked onto MacConkey agar (Condalab, Madrid, Spain) with and without cefotaxime (1 mg/L) (cefotaxime sodium salt; Acros Organics, Geel, Belgium) and incubated aerobically for 24 h at 37 °C. Cefotaxime-supplemented MacConkey agar served as a preliminary screen for isolates with reduced susceptibility to third-generation cephalosporins. Presumptive E. coli colonies were subcultured on Eosin Methylene Blue (EMB) agar (Liofilchem s.r.l., Teramo, Italy) and incubated for 24 h at 37 °C. Confirmation was performed using API 20E biochemical identification strips (BioMérieux, Marcy-l’Étoile, France) according to the manufacturer’s instructions. Confirmed E. coli isolates were preserved as stub cultures in nutrient agar (Condalab, Madrid, Spain) in microtubes for subsequent testing.

2.3. Antimicrobial Susceptibility Testing (AST)

Antimicrobial susceptibility was assessed using the disk diffusion method, following the Clinical and Laboratory Standards Institute (CLSI) and European Committee on Antimicrobial Susceptibility Testing (EUCAST) recommendations [15,16]. Stub-preserved isolates were revitalized on nutrient agar (Condalab, Madrid, Spain). Colonies were suspended in sterile saline and adjusted to 0.5 McFarland (BioMérieux, Marcy-l’Étoile, France; 1.5 × 10⁸ CFU/mL). Suspensions were inoculated evenly onto Mueller–Hinton agar (Condalab, Madrid, Spain) using sterile swabs. After a brief drying period, antibiotic disks (Liofilchem s.r.l., Teramo, Italy) were applied, and plates were incubated at 37 °C for 24 h. Inhibition zones were measured (mm) and interpreted according to CLSI recommendations [15] for all agents except tigecycline, which was interpreted using EUCAST criteria [16]. Isolates were classified as susceptible (S), intermediate (I), resistant (R), or susceptible dose-dependent (SDD). For cefepime, isolates were classified as SDD when the inhibition zone diameter was 19–24 mm. Multidrug resistance (MDR) was defined as resistance to at least one antimicrobial agent in three or more antimicrobial classes [17].

The following antibiotic disks and potencies were used: ampicillin (AMP, 10 µg), amoxicillin/clavulanic acid (AMC, 20 µg + 10 µg), cefepime (FEP, 30 µg), cefoxitin (FOX, 30 µg), cefotaxime (CTX, 30 µg), ceftazidime (CAZ, 30 µg), cefotaxime/clavulanic acid (CTL, 30 µg + 10 µg), ceftazidime/clavulanic acid (CAL, 30 µg + 10 µg), meropenem (MRP, 10 µg), imipenem (IMI, 10 µg), ertapenem (ETP, 10 µg), ciprofloxacin (CIP, 5 µg), azithromycin (AZM, 15 µg), tetracycline (TET, 30 µg), and tigecycline (TGC, 15 µg). The recommended approach for assessing colistin susceptibility is broth microdilution, in line with CLSI and EUCAST recommendations [15,16]. Colistin MICs were determined using Sensititre™ EUVSEC plates (TREK Diagnostic Systems, East Grinstead, United Kingdom), which include colistin alongside a fixed panel of additional antimicrobials. E. coli ATCC 25922 and E. coli NCTC 13846 were used for quality control.

2.4. Identification of ESBL-Producing E. coli

Phenotypic confirmation of ESBL-producing E. coli was performed using the Combination Disk Test (CDT) on isolates that grew on MacConkey agar supplemented with cefotaxime (1 mg/L). Standardized inocula (0.5 McFarland; BioMérieux, Marcy-l’Étoile, France; 1.5 × 10⁸ CFU/mL) were inoculated onto Mueller–Hinton agar (Condalab, Madrid, Spain). Disks containing cefotaxime and ceftazidime, as well as disks containing cefotaxime/clavulanic acid and ceftazidime/clavulanic acid (Liofilchem s.r.l., Italy), were applied to the plates, which were then incubated for 18 h at 37 °C. The CDT was considered positive when the inhibition zone diameter increased by ≥5 mm for cefotaxime/clavulanic acid or ceftazidime/clavulanic acid compared with the corresponding cephalosporin disk without clavulanic acid [15]. E. coli ATCC 25922 was used for quality control.

2.5. Bacterial DNA Extraction

Bacterial DNA was prepared using the boiling method. A single E. coli colony was inoculated into 5 mL BPW (Condalab, Madrid, Spain) and incubated overnight at 37 °C. One milliliter of culture was centrifuged at 14,000 rpm for 5 min. The supernatant was discarded, and the pellet resuspended in 300 µL sterile TE buffer (10 mM Tris, 0.1 mM EDTA, pH 8) (Sigma-Aldrich, Buchs, Switzerland). Suspensions were heated at 99 °C for 10 min, cooled on ice for 3 min, vortexed, centrifuged at 14,000 rpm for 2 min, and cooled again on ice. A 100 µL aliquot of supernatant (containing chromosomal and plasmid DNA) was used directly for PCR assays or stored at −20 °C until analysis.

2.6. PCR/qPCR Detection of Antimicrobial Resistance Genes

Targeted molecular screening was performed for β-lactamase genes (bla_TEM, bla_SHV, bla_CTX-M, bla_CMY), carbapenemase genes (bla_KPC, bla_NDM, bla_GES, bla_OXA-48), and plasmid-mediated colistin resistance genes (mcr-1 to mcr-9). bla_VIM was not included in the primer panel because no carbapenem non-susceptibility was observed phenotypically in the studied isolates, and the molecular screening was primarily designed to characterize ESBL/pAmpC and selected carbapenemase determinants. Primer/probe sequences used for AMR gene detection are provided in

Table 1. Sequences of specific primers and fluorescent probes are used to detect individual AMR genes.

Target Gene	Primer/Probe Name	Sequence (5′-3′)	Source
blaTEM	TEM_fwd.	GCATCTTACGGATGGCATGA	[18]
	TEM_rev.	GTCCTCCGATCGTTGTCAGAA
	TEM_probe	6-FAM-CAGTGCTGCCATAACCATGAGTGA-BHQ1
blaSHV	SHV_fwd.	TCCCATGATGAGCACCTTTAAA
	SHV_rev.	TCCTGCTGGCGATAGTGGAT
	SHV_probe	Cy5-TGCCGGTGACGAACAGCTGGAG-BBQ650
blaCTX-M	CTX_Afwd.	CGGGCRATGGCGCARAC
	CTX_A_rev.	TGCRCCGGTSGTATTGCC
	CTX_A_probe	VIC-CCARCGGGCGCAGYTGGTGAC-BHQ1
	CTX_B_fwd.	ACCGAGCCSACGCTCAA
	CTX-B_rev.	CCGCTGCCGGTTTTATC
	CTX_B_probe	Quasar705-CCCGCGYGATACCACCACGC-BHQ1
blaCMY	CMY_fwd.	GGCAAACAGTGGCAGGGTAT
	CMY_rev.	AATGCGGCTTTATCCCTAACG
	CMY_probe	ROX-CCTACCGCTGCAGATCCCCGATG-BHQ2
blaGES	GES_fwd.	CGGTTTCTAGCATCGGGACACAT	[19]
	GES_rev.	CCGCCATAGAGGACTTTAGCMACAG
	GES_probe	ATTO700-CGACCTCAGAGATACAACTACGCCTATTGC-DDQ1
blaKPC	KPC_fwd.	TGCAGAGCCCAGTGTCAGTTT	[20]
	KPC_rev.	CGCTCTATCGGCGATACCA
blaNDM	NDM_fwd.	CATTAGCCGCTGCATTGATG
	NDM_rev.	GTCGCCAGTTTCCATTTGCT
	NDM_probe	6-FAM-CATGCCCGGTGAAATCCGCC-BHQ1
blaOXA-48	OXA-48_fwd.	GCGTGGTTAAGGATGAACAC
	OXA_48_rev.	CATCAAGTTCAACCCAACCG
	OXA_48_probe	ROX-AGCCATGCTGACCGAAGCCAATG-BHQ2
mcr-1	mcr-1_fwd.	AGTCCGTTTGTTCTTGTGGC	[19]
	mcr-1_rev.	AGATCCTTGGTCTCGGCTTG
mcr-2	mcr-2_fwd.	CAAGTGTGTTGGTCGCAGTT
	mcr-2_rev.	TCTAGCCCGACAAGCATACC
mcr-3	mcr-3_fwd.	AAATAAAAATTGTTCCGCTTATG
	mcr-3_rev.	AATGGAGATCCCCGTTTTT
mcr-4	mcr-4_fwd.	TCACTTTCATCACTGCGTTG
	mcr-4_rev.	TTGGTCCATGACTACCAATG
mcr-5	mcr-5_fwd.	ATGCGGTTGTCTGCATTTATC
	mcr-5_rev.	TCATTGTGGTTGTCCTTTTCTG
mcr-6	mcr-6_fwd.	AGCTATGTCAATCCCGTGAT	[20]
	mcr-6_rev.	ATTGGCTAGGTTGTCAATC
mcr-7	mcr-7_fwd.	GCCCTTCTTTTCGTTGTT
	mcr-7_rev.	GGTTGGTCTCTTTCTCGT
mcr-8	mcr-8_fwd.	TCAACAATTCTACAAAGCGTG
	mcr-8_rev.	AATGCTGCGCGAATGAAG
mcr-9	mcr-9_fwd.	TTCCCTTTGTTCTGGTTG
	mcr-9_rev.	GCAGGTAATAAGTCGGTC

. Appropriate quality-control measures were included in each PCR run. Each PCR run included an NTC (PCR-grade water) and an extraction blank control (per extraction series) to monitor potential contamination. All reactions were performed in duplicate, and only concordant duplicate results were considered for reporting. For the reported PCR-positive findings, Ct values were <25, while no amplification was observed in the NTC or extraction blank control. Gene-specific positive controls were not available at the time of testing.

2.7. Multiplex Real-Time PCR for bla_TEM, bla_SHV and bla_CTX-M

Multiplex real-time PCR was performed on a QuantStudio™ 5 Real-Time PCR System, and results were analyzed using QuantStudio™ Design and Analysis Software v1.5.1, following an adapted protocol of Roschanski et al. (2014) [18]. The final reaction volume was 25 µL and contained 5.5 µL of 5× QuantiFast Pathogen Master Mix pre-prepared with ROX passive dye (50× ROX Dye Solution) (Qiagen, Hilden, Germany), 1 µL of each forward and reverse primer (10 pmol), 0.5 µL TEM probe (5 pmol), 0.5 µL of each of the remaining probes (10 pmol) (Table 1), 4.5 µL nuclease-free PCR water, and 5 µL of DNA template. Thermocycling comprised 95 °C for 5 min, followed by 45 cycles of 95 °C for 15 s and 60 °C for 30 s (fluorescence acquired during annealing/extension). Full cycling details are provided in Supplementary Table S4.

2.8. Multiplex Real-Time PCR for bla_CMY, bla_NDM and bla_GES

Multiplex assays for bla_CMY, bla_NDM and bla_GES were performed on a Mic real-time PCR cycler (Bio Molecular Systems, Upper Coomera, Australia) with analysis in micPCR software (v2.8.13), using adapted protocols [18,19]. Reactions (25 µL) contained 5 µL of 5× QuantiFast Pathogen Master Mix (Qiagen, Hilden, Germany), 1 µL of each forward and reverse primer (10 pmol), 0.5 µL of each TaqMan probe (10 pmol) (Table 1), 7.5 µL nuclease-free PCR water, and 5 µL of DNA template. Thermocycling conditions followed the same profile (Supplementary Table S4).

2.9. Real-Time PCR for bla_OXA-48

Detection of bla_OXA-48 was performed on the Mic real-time PCR cycler using an adapted protocol [19]. The 25 µL reaction contained 5 µL of 5× QuantiFast Pathogen Master Mix (Qiagen, Germany), 1 µL of each forward and reverse primer (10 pmol), 0.5 µL OXA-48 TaqMan probe (10 pmol) (Table 1), 12.5 µL nuclease-free PCR water, and 5 µL of DNA template. Thermocycling conditions followed the same profile (Supplementary Table S4).

2.10. EvaGreen-Based Real-Time PCR for bla_KPC and Colistin Resistance Genes (mcr-1 to mcr-9)

Specific detection of bla_KPC and colistin resistance genes (mcr-1 to mcr-9) was performed using a modified approach [20,21]. DNA amplification was conducted on the Mic real-time PCR cycler using 2× Fast EvaGreen qPCR MasterMix (Biotium, Fremont, CA, USA) and gene-specific primers (Table 1). EvaGreen assays were run in a 20 µL reaction volume (15 µL master mix + 5 µL DNA template), using 2× Fast EvaGreen qPCR MasterMix and gene-specific primers. Cycling included 95 °C activation, 45 cycles of denaturation/annealing/extension, and melt curve analysis. Full reaction setup and cycling parameters are provided in Supplementary Tables S5 and S6.

2.11. Data Handling and Statistical Analysis

Data were managed and analyzed using Stata/SE 15 (StataCorp, College Station, TX, USA). Results were summarized using descriptive statistics and presented as counts and proportions; where appropriate, 95% confidence intervals (95% CI) were reported. Group comparisons for categorical variables (e.g., phenotypic resistance, ESBL status, MDR status, and presence of resistance genes) were performed using Pearson’s chi-square test or Fisher’s exact test when expected cell counts were small. All tests were two-sided, and p ≤ 0.05 was considered statistically significant.

3. Results

3.1. Recovery of E. coli and ESBL Phenotype Across Poultry Compartments

A total of 333 samples were examined across three poultry production compartments, including a parent-breeder farm (n = 95), commercial broiler farms (n = 163), and a hatchery (n = 75). Overall, 99 E. coli isolates were recovered (29.7%, 99/333), comprising 28/95 (29.5%) isolates from the parent-breeder compartment, 51/163 (31.3%) from broiler farms, and 20/75 (26.7%) from hatchery samples (

Table 2. Isolation of E. coli and distribution of ESBL and non-ESBL isolates by poultry production compartment.

Sample Origin	Number of E. coli Isolates (%)			Total (%)
	ESBL n (%)	Non-ESBL n (%)	Total n (%)
Parent-breeder farm	18 (64.3%)	10 (35.7%)	28 (29.5%)	95 (28.5%)
Broiler farms	28 (54.9%)	23 (45.1%)	51 (31.3%)	163 (48.9%)
Hatchery	6 (30%)	14 (70%)	20 (26.7%)	75 (22.5%)
Total	52 (52.5%)	47 (47.5%)	99 (29.7%)	333 (100%)

). Phenotypic ESBL status was determined for all isolates, yielding 52 ESBL-producing isolates (52.5%, 52/99) and 47 non-ESBL isolates (47.5%, 47/99). The proportion of ESBL isolates varied numerically across compartments (parent-breeder: 18/28, 64.3%; broilers: 28/51, 54.9%; hatchery: 6/20, 30.0%) but did not reach statistical significance (Pearson’s χ² test, p = 0.057).

3.2. Phenotypic Antimicrobial Susceptibility Testing

3.2.1. Disk Diffusion Susceptibility Profiles

Disk diffusion testing indicated frequent resistance to multiple antimicrobial classes among poultry- and hatchery-derived E. coli isolates. Resistance to ampicillin was common in all compartments (parent-breeder: 92.9%, 26/28; broilers: 76.5%, 39/51; hatchery: 60.0%, 12/20). Tetracycline resistance was also prevalent (parent-breeder: 67.9%, 19/28; broilers: 62.7%, 32/51; hatchery: 35.0%, 7/20), as was ciprofloxacin resistance (parent-breeder: 28.6%, 8/28; broilers: 49.0%, 25/51; hatchery: 25.0%, 5/20). Resistance to third-generation cephalosporins was detected, including cefotaxime (parent-breeder: 35.7%, 10/28; broilers: 35.3%, 18/51; hatchery: 30.0%, 6/20) and ceftazidime (broilers: 33.3%, 17/51; hatchery: 45.0%, 9/20). No resistance to carbapenems (meropenem, ertapenem, imipenem) was recorded by disk diffusion in any poultry production compartment, and all isolates were classified as susceptible to tigecycline. A summary of resistance prevalence by disk diffusion across compartments is provided in

Table 3. Phenotypic resistance of E. coli isolates by disk diffusion across poultry production compartments (resistant isolates only).

Antimicrobials	Parent-Breeder Farms (n = 28)	Broiler Farms (n = 51)	Hatchery Samples (n = 20)
	No. (%) of Resistant E. coli Isolates
Ampicillin	26 (92.9%)	39 (76.5%)	12 (60%)
Amoxicillin/clavulanic acid	4 (14.3%)	7 (13.7%)	6 (30%)
Cefepime	3 (10.7%)	12 (23.5%)	1 (5%)
Cefoxitin	7 (25%)	12 (23.5%)	5 (25%)
Cefotaxime	10 (35.7%)	18 (35.3%)	6 (30%)
Ceftazidime	0%	17 (33.3%)	9 (45%)
Meropenem	0%	0%	0%
Ertapenem	0%	0%	0%
Imipenem	0%	0%	0%
Ciprofloxacin	8 (28.6%)	25 (49%)	5 (25%)
Azithromycin	0%	4 (7.8%)	0%
Tetracycline	19 (67.9%)	32 (62.7%)	7 (35%)
Tigecycline a	0%	0%	0%

^a Data are presented as the number of resistant isolates (%). Susceptibility categories were interpreted according to CLSI [15] (all agents) and EUCAST [16] (tigecycline).

. Detailed S/I/R/SDD distributions stratified by ESBL phenotype are provided in Supplementary Tables S1–S3.

3.2.2. Broth Microdilution (EUVSEC Panel)

Broth microdilution (Sensititre™ EUVSEC) was used to determine categorical susceptibility outcomes for colistin and additional antimicrobials in E. coli isolates from parent-breeder flocks, broiler flocks, and hatchery samples, stratified by ESBL phenotype (Figure 1). Overall, no colistin-resistant isolates were detected across any compartment or ESBL category (0% resistant in all groups). Likewise, no meropenem-resistant isolates were observed (0% resistant in all groups), and tigecycline resistance was not detected (0% resistant in all groups).

In contrast, resistance to several non-last-line agents was frequent. Ampicillin resistance was consistently high, reaching 100% in parent-breeder ESBL isolates (18/18) and hatchery ESBL isolates (6/6), and remaining substantial in broiler isolates (e.g., 82.1% in broiler ESBL, 23/28). Chloramphenicol resistance was very common, including 100% resistance in parent-breeder isolates (ESBL 18/18; non-ESBL 10/10), and >90% in broiler isolates (ESBL 92.9%, 26/28; non-ESBL 91.3%, 21/23); hatchery non-ESBL isolates were uniformly resistant (14/14). Gentamicin resistance followed a similar pattern (often ≥90–100% across strata). Resistance to extended-spectrum cephalosporins was mainly concentrated among ESBL isolates, including cefotaxime resistance of 55.6% (10/18) in parent-breeder ESBL isolates, 64.3% (18/28) in broiler ESBL isolates, and 100% (6/6) in hatchery ESBL isolates, while non-ESBL isolates were largely non-resistant to cefotaxime (0% resistant across all non-ESBL groups, i.e., non-ESBL isolates in each production compartment). Fluoroquinolone resistance (ciprofloxacin) was also observed, with the highest proportion in broiler ESBL isolates (67.9%, 19/28). Figure 1 displays the percentage of resistant isolates (R) for each antimicrobial within each compartment and ESBL stratum. For colistin, categorical outcomes were interpreted as binary S/R (no intermediate category), consistent with the EUVSEC-based approach. In Figure 1, agents for which interpretive criteria were not applied in the EUVSEC analysis (sulfamethoxazole, azithromycin, and nalidixic acid) are shown as NA.

3.2.3. Multidrug Resistance Profile

Multidrug resistance (MDR), defined as resistance to at least one agent in three or more antimicrobial classes, was highly prevalent among poultry- and hatchery-derived E. coli isolates. Overall, 93/99 (93.9%) isolates fulfilled the MDR criterion, including 74/79 (93.7%) isolates from poultry farms and 19/20 (95.0%) isolates from hatchery samples (

Table 4. Distribution of multidrug resistance by number of resistant antimicrobial classes among poultry- and hatchery-derived E. coli isolates, stratified by ESBL phenotype.

No. of Resistant Antimicrobial Classes	Poultry Farms (n = 79)		Hatchery Samples (n = 20)
	ESBL (n = 46)	Non-ESBL (n = 33)	ESBL (n = 6)	Non-ESBL (n = 14)
0	1 (2.2%)	2 (6.1%)	0%	0%
1	1 (2.2%)	0%	0%	0%
2	0%	1 (3%)	0%	1 (7.1%)
3	6 (13%)	5 (15.2%)	1 (16.7%)	9 (64.3%)
4	22 (47.8%)	22 (66.7%)	2 (33.3%)	4 (28.6%)
5+	16 (34.8%)	3 (9.1%)	3 (50%)	0%
Multiresistant isolates (%) a	44 (95.7%)	30 (90.9%)	6 (100%)	13 (92.9%)

^a MDR is defined as resistance to ≥1 agent in ≥3 antimicrobial classes. β-lactams are discussed as subclasses (penicillins, cephalosporins/cephamycins, carbapenems).

). When stratified by ESBL phenotype, MDR was observed in 44/46 (95.7%) ESBL and 30/33 (90.9%) non-ESBL isolates from poultry farms, as well as in 6/6 (100%) ESBL and 13/14 (92.9%) non-ESBL isolates from the hatchery. Most isolates exhibited resistance to four antimicrobial classes, whereas resistance to five or more classes was particularly frequent among ESBL isolates (Table 4).

3.3. Molecular Detection of Antimicrobial Resistance Genes

3.3.1. Distribution of ESBL/pAmpC Genes Across Poultry Compartments

Overall, ESBL/pAmpC-associated genes (bla_TEM, bla_SHV, bla_CTX-M, bla_CMY, and their combinations) were detected in 91/99 (91.9%) E. coli isolates recovered from poultry production compartments. Among PCR-positive isolates, Ct values (min–max) were: bla_TEM, 16.1–22.3; bla_SHV, 14.4–19.7; bla_CTX-M, 17.8–23.0; and bla_CMY, 18.2–24.6. Gene detection frequencies were 25/28 (89.3%) in parent-breeder isolates, 49/51 (96.1%) in broiler isolates, and 17/20 (85.0%) in hatchery isolates. Considering ESBL phenotype, at least one ESBL/pAmpC-associated gene was detected in 51/52 (98.1%) ESBL isolates and 40/47 (85.1%) non-ESBL isolates. Across compartments, bla_TEM was the most frequently detected determinant, either alone or in combination with other genes. bla_SHV occurred sporadically, whereas bla_CTX-M and bla_CMY were less frequently detected as single targets but appeared in defined gene combinations. The distribution of individual genes and gene combinations, stratified by production compartment and ESBL phenotype, is summarized in Figure 2.

3.3.2. Carbapenemase and Plasmid-Mediated Colistin Resistance Genes

None of the poultry- and hatchery-derived E. coli isolates carried carbapenemase-associated genes (bla_KPC, bla_NDM, bla_GES, bla_OXA-48) or plasmid-mediated colistin resistance genes (mcr-1 to mcr-9) in the study dataset.

4. Discussion

In the present study, we investigated commensal E. coli across three interconnected segments of the poultry production chain in Bosnia and Herzegovina (parent-breeder flocks, broiler flocks, and hatchery-associated material), combining phenotypic antimicrobial susceptibility testing (including ESBL confirmation and MDR profiling) with targeted molecular screening of selected resistance determinants. ESBL phenotypes were detected across all examined production stages, accompanied by a high proportion of MDR isolates, indicating that resistant E. coli is present throughout the evaluated production environment rather than being confined to a single stage. Although ESBL proportions differed numerically between compartments, this difference did not reach statistical significance (p = 0.057), suggesting limited power for compartment-level contrasts and supporting cautious interpretation of stage-specific differences. The predominance of β-lactam resistance together with frequent co-resistance is in line with the established role of commensal E. coli as a reservoir of transferable antimicrobial resistance determinants within food-production systems [22,23]. At the same time, the absence of the investigated carbapenemase- and plasmid-mediated colistin-resistance genes suggests that these last-resort resistance determinants were not detected in this dataset, although this should be interpreted within the scope of the targeted gene panel and study design [23]. Collectively, these findings provide stage-specific evidence that ESBL/pAmpC-associated resistance is detected across the poultry chain and highlight the value of integrating phenotypic testing with molecular screening when interpreting AMR patterns in poultry-associated E. coli [23,24].

4.1. ESBL/pAmpC Burden and β-Lactam Resistance Determinants Across Production Stages

The high frequency of the ESBL phenotype among isolates from parent-breeder flocks (64.3%) and broiler flocks (54.9%), together with a lower frequency in hatchery-associated material (30.0%), indicates that β-lactam resistance represents a substantial burden across multiple production stages, although both intensity and resistance profiles vary by source. The predominance of β-lactam resistance determinants, i.e., most notably bla_TEM (alone or in combination), with detection of bla_CTX-M and/or bla_CMY in a subset of isolates, is consistent with the interpretation that antimicrobial resistance to penicillins and cephalosporins in this population is largely genetically mediated. In this study, the ESBL phenotype was defined using the clavulanate-based combination disk test, whereas bla_CMY was interpreted as a marker of plasmid-mediated AmpC. Because bla_TEM and bla_SHV encompass both narrow-spectrum and ESBL alleles, allele-level resolution (e.g., sequencing/WGS) would be required to attribute ESBL activity to specific variants with certainty. In addition, the absence of carbapenemase genes (bla_KPC, bla_NDM, bla_GES, bla_OXA-48) is consistent with patterns typically observed in poultry-related settings in Europe [7,23].

Compared with previously published findings from Bosnia and Herzegovina, our results complement the existing evidence documenting ESBL/pAmpC-positive E. coli in poultry production and poultry products [12,13]. Nevertheless, direct comparisons of prevalence should be interpreted cautiously, given differences in sample type, matrix (farm vs. product), sampling period, and methods used for phenotypic and genotypic confirmation. Even so, the repeated identification of ESBL/pAmpC-associated determinants across different segments of poultry production underscores that control of β-lactam resistance is relevant both at the level of primary production and downstream along the production chain [7,23,24].

From the broiler production perspective, it was noted that ESBL/pAmpC-producing Enterobacterales can be detected at multiple levels of broiler production, reflecting a combination of potential vertical and horizontal mechanisms sustaining these bacteria within integrated systems [9]. The evidence from the broiler production pyramid has demonstrated the presence of ESBL/AmpC-producing E. coli at different points along the chain [25], while longitudinal observations have further indicated a possible vertical component in the distribution of pAmpC-positive strains [26]. In our context, the comparatively higher ESBL proportion observed in parent-breeder flocks in our material could be compatible with an upstream reservoir or pressure. However, it cannot be interpreted as evidence of transmission without strain-level data, i.e., phylogenetics and/or WGS approaches.

The hatchery warrants specific attention. Although the ESBL phenotype was less frequent in our hatchery-associated material than on farms, hatcheries have been previously described as potential points of amplification and redistribution of resistant bacteria, including E. coli, which may be relevant for targeted biosecurity interventions [27]. Therefore, a lower ESBL frequency in this segment does not exclude epidemiological importance but may instead reflect sample-matrix characteristics (hatchery waste), temporal dynamics of contamination, or differences in selection pressure relative to farm environments.

The presence of bla_TEM, bla_CTX-M and/or bla_CMY aligns with broader patterns reported in poultry, where ESBL and pAmpC mechanisms frequently dominate cephalosporin resistance, although with variability in prevailing genes across regions, time periods, and production systems [28]. Within ESBL groups, enzymes of the CTX-M family are recognized as globally important, with marked expansion across hosts and ecological niches [29], which is relevant when interpreting bla_CTX-M detection in a subset of our isolates. Conversely, the absence of carbapenemase genes in our samples is in line with European surveillance reports, where carbapenem resistance in indicator bacteria from food animals and food is typically described as rare [7].

4.2. Multidrug Antimicrobial Resistance and Co-Resistance Patterns Across the Poultry Chain

One of the key findings of this study is the consistently high prevalence of multidrug resistance (MDR) across all examined production stages. Using the accepted interim definition (antimicrobial resistance to at least one antimicrobial agent in three or more classes) [29], MDR was detected in more than 90% of E. coli isolates from parent-breeder flocks, broiler flocks, and hatchery-associated material. Such uniformly high MDR levels indicate that antimicrobial resistance in poultry-associated commensal E. coli in Bosnia and Herzegovina is not limited to individual compounds or classes but rather reflects stable multiclass antimicrobial resistance phenotypes within the production environment.

The observed MDR patterns are compatible with established knowledge on the genetic organization of antimicrobial resistance in Enterobacterales from food-producing animals. AMR determinants are frequently located on mobile genetic elements, particularly plasmids, where genes conferring antimicrobial resistance to different antimicrobial classes may be physically linked [30,31]. Under these conditions, selection pressure exerted by one antimicrobial class may maintain antimicrobial resistance to others through co-selection, even in the absence of direct exposure to all classes involved. In the present study, the frequent detection of β-lactam resistance determinants (predominantly bla_TEM, with bla_CTX-M and/or bla_CMY in a subset of isolates; see Section 4.1), together with the very high MDR prevalence, supports the interpretation that co-resistance plays an important role in shaping the antimicrobial resistance profiles observed in poultry-associated E. coli. However, as plasmid types and full antimicrobial resistance gene contexts were not investigated, conclusions regarding specific genetic structures cannot be drawn.

MDR in commensal E. coli should also be viewed in the context of production dynamics. Studies conducted in broiler systems have demonstrated that antimicrobial resistance profiles may change during the production cycle, influenced by antimicrobial exposure, management practices, and within-flock microbial interactions [32]. In addition, broader evaluations of antimicrobial use in broiler production have consistently associated antimicrobial application with the selection and persistence of resistant E. coli, often extending beyond the antimicrobial class directly used [33]. In this study, the similarly high MDR proportions observed in farm-derived isolates and hatchery-associated isolates suggest that multiple ecological niches within the poultry production chain may contribute to the maintenance of MDR populations, rather than MDR being generated at a single, isolated point.

From a practical perspective, the relevance of these findings lies in the fact that MDR profiles frequently include antimicrobial classes regarded as critically or highly important for human and veterinary medicine [34,35]. Therefore, the high MDR burden observed in this study underscores the need for integrated control strategies that address antimicrobial use, biosecurity, and hygiene across the entire poultry production chain. Further investigations incorporating strain-level approaches, such as whole-genome sequencing combined with plasmid analysis, would be essential to determine whether the observed MDR patterns are driven primarily by clonal expansion, horizontal gene transfer, or their combination, and to better contextualize the phenotypic co-resistance documented here [30,31].

4.3. Antimicrobial Resistance Beyond ESBL-Associated β-Lactams and Multidrug Resistance Patterns Across Production Stages

Consistent with the stage-stratified susceptibility profiles (Table 3), antimicrobial resistance was not limited to cephalosporins and ESBL-related phenotypes, but extended to several non-β-lactam classes that are frequently used in poultry production and/or considered highly relevant from a One Health perspective. Across all three segments, antimicrobial resistance to ampicillin remained very common (60.0–92.9%), while tetracycline resistance was also frequent on farms (67.9% in parent-breeder isolates; 62.7% in broiler isolates) and lower in hatchery-associated isolates (35.0%). Fluoroquinolone resistance showed a clear source-dependent pattern, peaking in broiler isolates (ciprofloxacin 49.0%) compared with parent-breeder (28.6%) and hatchery-associated isolates (25.0%). In contrast, azithromycin resistance was uncommon (detected only among broiler isolates, 7.8%), and no antimicrobial resistance to tigecycline or the tested carbapenems was detected in any production stage (Table 3). These findings support the interpretation that, within this dataset, the main selective ‘pressure footprint’ is expressed through older β-lactams and commonly used classes such as tetracyclines and fluoroquinolones, whereas last-resort agents for human medicine (e.g., carbapenems; and, in our panel, tigecycline) remained fully active in this dataset [23,34]. From a stewardship standpoint, the presence of substantial fluoroquinolone resistance deserves particular attention, given the status of this class among critically important antimicrobials for human medicine [34], alongside its relevance in veterinary medicine frameworks [35].

When antimicrobial resistance was evaluated as a multi-class phenotype (Table 4), multidrug resistance was nearly ubiquitous in both farm-derived isolates and hatchery-associated isolates. MDR proportions reached 93.7% among poultry-farm isolates overall and 95.0% among hatchery isolates overall. Notably, the distribution of resistant class counts differed by ESBL status. On farms, ESBL-positive isolates more frequently accumulated resistance to five or more antimicrobial classes (34.8%) compared with non-ESBL isolates (9.1%), whereas non-ESBL farm isolates were more often concentrated at four resistant classes (66.7%) (Table 4). A similar gradient was observed in hatchery-associated isolates: half of ESBL-positive hatchery isolates displayed antimicrobial resistance to ≥5 classes, while none of the non-ESBL hatchery isolates reached that category (Table 4). In practical terms, this pattern indicates that ESBL positivity in our material is not an isolated β-lactam trait, but is commonly embedded in broader MDR profiles. This observation is consistent with the well-described linkage of multiple antimicrobial resistance determinants on mobile genetic elements in Enterobacteriaceae [30,31]. Even without resolving plasmid backbones or complete resistomes, the clear shift toward higher ‘class counts’ among ESBL-positive isolates (Table 4), together with frequent antimicrobial resistance to tetracycline and ciprofloxacin (Table 3), supports the interpretation that co-resistance and co-selection are likely important in maintaining these phenotypes within poultry-associated E. coli populations [30,31].

The overall antimicrobial resistance pattern observed in our study, i.e., high antimicrobial resistance to ampicillin and tetracycline coupled with variable, but sometimes substantial, fluoroquinolone resistance, aligns with trends commonly reported for commensal/indicator E. coli in broiler production, although absolute levels vary considerably across countries, production systems, and sampling designs [7,32,33]. Comparable antimicrobial resistance profiles in commensal E. coli from broiler flocks have also been reported at the national level in Bosnia and Herzegovina, supporting the relevance of these findings in the local production context [36]. Regional data from Bosnia and Herzegovina also support the presence of resistant E. coli in poultry-associated matrices [12,13], while studies from other settings have similarly highlighted that ESBL/pAmpC-positive populations in broiler systems often co-occur with antimicrobial resistance beyond β-lactams, contributing to high MDR burdens at the farm level and, in some circumstances, within hatchery environments [27,37,38]. Importantly, such cross-study comparisons should be interpreted cautiously because they are influenced by differences in the sampled matrix (farm feces vs. hatchery material vs. carcass/skin), age and production stage, antimicrobial exposure history, and the exact interpretive criteria applied [7,32,33]. Nevertheless, taken together, the very high MDR frequency in our dataset (Table 4), combined with consistent resistance to commonly used classes (Table 3), suggests that interventions aimed solely at cephalosporin/ESBL control are unlikely to substantially reduce the broader antimicrobial resistance burden without parallel improvements in antimicrobial stewardship and biosecurity across the production continuum [33,34]. The hatchery segment merits continued attention in this regard: even with a lower ESBL proportion than on farms (Section 4.1), hatcheries have been described as potential reservoirs and redistribution points for resistant E. coli, which supports targeted hygiene and monitoring strategies tailored to hatchery workflows and contamination dynamics [27]. Environmental persistence is particularly relevant when antimicrobial resistance is accompanied by broad co-resistance patterns. Studies from poultry farm environments have described ESBL-producing E. coli in different matrices and at variable loads, suggesting that the ‘production environment’ itself may act as a reservoir that can complicate control efforts if interventions focus only on a single point (e.g., only farms or only hatcheries) [38]. In addition, cross-border data from Belgian and Dutch broiler and pig farms have addressed combined ESBL- and fluoroquinolone resistance, including phenotypes involving critically important drug classes, underscoring why continued monitoring remains justified even when last-resort resistance determinants are not detected in a given dataset [39]. Taken together with European surveillance reporting that carbapenem resistance in indicator bacteria from food animals and food is typically rare, our finding of no phenotypic carbapenem resistance and no carbapenemase genes aligns with the wider European picture while still pointing to a substantial ESBL/MDR burden that is relevant for stewardship and biosecurity across the production chain [7].

Several limitations should be acknowledged. First, the study relied on commensal/indicator isolates obtained from a limited number of compartments and sampling units, which may constrain the generalizability of stage-specific comparisons. Second, ESBL screening and phenotypic confirmation were performed using culture-based approaches and targeted molecular assays; however, allele-level resolution and plasmid context were not assessed, and whole-genome sequencing would be required to distinguish clonal spread from horizontal gene transfer and to characterize the resistome more comprehensively. In addition, gene-specific positive controls were not available for the PCR assays. PCR results are therefore reported as detection findings and interpreted with appropriate caution. Finally, the cross-sectional design and the absence of longitudinal sampling limit inference on temporal dynamics and directionality of dissemination between farms and hatchery-associated material.

5. Conclusions

This study indicates that the poultry production chain in Bosnia and Herzegovina, including both farms and hatchery-associated material, harbors a high background burden of ESBL/MDR commensal/indicator E. coli. The overall similarity of antimicrobial resistance patterns between compartments, together with the lack of statistically supported stage-specific differences, suggests that compartment-focused actions alone are unlikely to be sufficient; instead, stewardship and biosecurity should be implemented consistently across the entire chain, with explicit attention paid to hatchery hygiene and management. From a risk perspective, the absence of detected carbapenemase genes and plasmid-mediated colistin resistance genes in the screened panel is reassuring, yet it should not dilute the practical significance of the extensive MDR/ESBL burden observed. Overall, the findings support continued integrated phenotypic–molecular surveillance to guide targeted interventions and to enable early detection of emerging antimicrobial resistance threats.