Phenotypic and Genotypic Characterization of Colistin, ESBL, and Multidrug Resistance in Escherichia coli Across the Broiler Production Chain in Karnataka, India

Abstract

The emergence of antimicrobial resistance (AMR) across the broiler production chain holds significant economic, animal, and public health implications. This study investigated phenotypic resistance to 13 antimicrobials and the presence of 35 antimicrobial resistance genes (ARGs) in Escherichia coli (n = 291) isolated across three broiler production chains (broiler breeder farms, hatcheries, commercial broiler farms, and retail meat shops). An extremely high phenotypic resistance (>70%) to doxycycline, ciprofloxacin, and cefpodoxime, and very high resistance (50–70%) to ampicillin, cefotaxime, gentamicin, and ceftazidime was observed. In addition, 97% of isolates were multidrug-resistant (resistant to ≥1 drug in ≥3 antimicrobial classes), 42% were extended-spectrum beta-lactamase (ESBL) producers, 65% were resistant to third-generation cephalosporins (3GCR), and 21% were resistant to colistin. The Poisson regression model revealed no significant difference in AMR among broiler production stages, except for colistin. Among 35 ARGs tested, 24 (67%) were detected at least once. The most prevalent were tetA, bla_TEM, qnrB, qnrS, and aac(6′)-Ib-cr, while qnrD, sul2, bla_OXA, and bla_CTX-M were detected at lower levels (1–7%). All five tested mcr genes (mcr-1 to mcr-5) were identified in commercial farms and retail shops. No bla_NDM, tetB, tetC, tetD, tetM, qnrC, aac(3)-IIa (aacC2), aph(3)-IIa (aphA2), or aac(6′)-Ib genes were found. A strong correlation was observed between AMR phenotypes and ARGs. High AMR among E. coli in broiler production poses significant One Health risks, with widespread MDR, ESBL production, and resistance to critically important antimicrobials. Prudent antimicrobial use, enhanced surveillance and education, farm biosecurity, and One Health strategies are crucial in mitigating these threats.

1. Introduction

India is one of the world’s largest poultry producers, with rapidly growing broiler and layer production driven by increasing demand for affordable animal protein. The country is the third-largest egg producer and the fifth-largest broiler producer worldwide [1]. Broiler chickens are raised for meat production on commercial poultry farms and reach market weight within 6 weeks. In India, the broiler production chain includes broiler breeders, hatcheries, commercial broiler farms, slaughterhouses, wholesalers, and retail meat shops [2]. The intensive farming practices in India, combined with limited on-farm biosecurity and significant economic losses due to bacterial poultry diseases, have led to extensive use of antibiotics for prophylactic, growth-promoting, and therapeutic purposes, potentially driving the selection of antimicrobial-resistant (AMR) bacteria [3,4].

Escherichia coli are a commensal organism commonly found in the intestinal tract of humans, poultry, and other warm-blooded animals. However, certain strains of E. coli can be pathogenic, causing disease, and are a frequent cause of bacterial infections in both humans and animals [5]. In poultry, colibacillosis, caused by avian pathogenic E. coli (APEC), results in several infections, including omphalitis, peritonitis, septicemia, and salpingitis [6]. Therefore, it leads to significant economic losses in the poultry industry worldwide as it reduces egg production in broiler breeders, increases mortality, and could be vertically transmitted to broiler chicks through eggs. It can also cause high first-week mortality in chicks due to horizontal transmission at hatcheries, with the added concern of vertical transmission of antibiotic resistance genes [7,8]. Colibacillosis is the third most commonly reported infection in broilers in India [3]. Further, commensal E. coli can be an indicator for the selection pressure of antimicrobial use (AMU) in both clinical and agricultural settings [9] due to its ability to acquire resistance genes, making it an ideal sentinel organism for monitoring AMR [5,10,11].

Highest Priority Critically Important Antimicrobials (HP-CIAs) are a subset of antibiotics identified by the World Health Organization (WHO) as essential for human health [12]. These antimicrobials, including fluoroquinolones, third- and fourth-generation cephalosporins, and colistin, are vital for treating severe bacterial infections, especially in cases of multidrug-resistant (MDR; resistant to at least one agent in ≥3 antimicrobial classes) [13] pathogens. Preserving their efficacy for treatment is crucial to combat AMR, as their loss would significantly affect public health [14].

The emergence of AMR in animals and humans is a global health issue, and one significant contributing factor to its rise in animal production systems is inappropriate AMU [15]. In recent years, an increase in AMR and MDR E. coli has been described [13,16]. A growing concern is the emergence of extended-spectrum beta-lactamase (ESBL)-producing and colistin-resistant E. coli in humans and animals, which has been reported worldwide, posing a significant global health challenge [17,18,19,20]. Along with other antimicrobials, colistin has been commonly administered to animals, especially in low- and middle-income countries [21]. Colistin is the last-resort antimicrobial for MDR pathogens in humans, targeting only Gram-negative pathogens with a narrower spectrum [22]. Due to the emergence and detection of mobile colistin-resistant genes (mcr) in E. coli from animals [23,24,25], several countries stopped the use of colistin in food animals. In India, colistin use was banned on 19 July 2019 [26]. However, there are still reports of colistin-resistant bacteria in retail chicken samples, on poultry farms, and in processing plants [20,27]. Retail meat shop outlets in India are a unique aspect of the poultry supply chain. These outlets typically receive live broilers daily, house them in small cages on-site, and offer a “live-to-table” service by slaughtering, processing, and selling the birds directly to customers. This process is often conducted in small, localized, unsanitary settings, usually in front of the buyer, to ensure freshness and to meet cultural preferences, and accounts for 95% of chickens slaughtered by retailers [28,29].

AMR in E. coli isolated from broilers has been extensively studied [30]. However, most research focuses on one or two segments of the broiler production chain, predominantly on commercial broiler farms or slaughterhouses. There are limited investigations of AMR phenotypes and their associated genes across the entire broiler production chain and its environment, from the beginning at broiler breeder farms to hatcheries, commercial broiler farms, and retail meat shops while targeting both AMR phenotypes and associated AMR genes.

This study examined the entire broiler production chain from the beginning of production until it reached the consumers at the retail meat shops, investigating emerging AMR phenotypes and genes.

2. Materials and Methods

2.1. Study Location and Sampling Sources

Samples (n = 332) were collected within a 40-mile radius of Bengaluru city and the Kolar district of Karnataka, in southern India, a key region for broiler production. The sampling frame included the broiler production chain and its environment, comprising three broiler breeder farms (BBFs, n = 81), three hatcheries (n = 36), three commercial broiler farms (CBF, n = 160) sampled at the beginning, middle, and end of the crop cycle (days 1, 18, and 36), and three retail meat shops (RMSs, n = 57). These samples were from three different broiler chicken integrators (a company or entity that manages the entire broiler production chain, including BBFs, hatcheries, feed mills, CBFS, processing, and marketing). In each integration, the sample batch was tracked across the entire production chain, from BBFs to the hatcheries receiving eggs from those breeders, then to the CBFs that raised these chicks, and finally to the RMSs that received those broilers, ensuring that samples at each stage corresponded to the same broiler batch. At each stage, a variety of sample types were collected, including water and feed samples, cloacal and fecal swabs, egg surfaces, yolk swabs, environmental swabs (boot socks, incubators, air tunnels, and hatchers), hand swabs from hatchery workers, carcass and ileal/caecal contents, and swabs from retail equipment surfaces (cutting boards and knives). Details of the sample types and numbers at each stage are provided in

Table 1. Samples collected from the broiler production chain.

No.	Types of Samples Collected	No. of Samples
Broiler breeder farms (BBFs)		81
1	Water from the water tank of the farm	9
2	Water from nipples from 30 nipples pooled into one/shed)	12
3	Feed sample from feed bags (10 bags pooled into one sample/shed)	12
4	Feed sample from feeders (From 30 areas/feeders pooled/shed)	12
5	Cloacal swabs (30 birds/shed and pooled)	12
6	Egg surface (30 eggs/shed and pooled)	12
7	Swabs from Egg Yolk (30 eggs/shed pooled)	8
8	Environmental soil sample (Boot socks): one pair per farm	4
Hatcheries		36
1	Swabs from egg setting room (10 swabs pooled to one per Hatchery)	4
2	Swabs from the Incubator (3 swabs from different areas of the incubator)	4
3	Swabs from air tunnels and fans of incubators	4
4	Swabs from hatchers	3
5	Swabs from Hatcher Egg Tray	3
6	Meconium	3
7	Yolk sac swabs of dead chicks (Ten dead chicks and samples were pooled into one/hatchery)	3
8	Hand swabs from hatchery workers	5
9	Boot socks from the hatchery floor	7
Commercial broiler farms, Day 1, Day 18, Day 36		160
1	Water from the water tank	25
2	Water from nipples/drinkers	25
3	Feed samples from feed bags	25
4	Feed feeders	25
5	Faecal swabs	25
6	Internal (inside the shed) environment samples using sterile boot socks	25
7	External (outside the shed) environment samples using sterile boot socks	10
Retail meat shops		57
1	Swabs from the surface of the cutting/chopping board (100 cm2)	6
2	Swabs from the cutter/knife	3
3	Meat rinsing water	3
4	Chicken carcasses (5 Carcasses/shop).	15
5	Ileal contents from five carcasses	15
6	Caecal contents from five carcasses	15
	Total number of samples	332

and Figure 1. Sample collections were performed at farms without significant outbreaks of bacterial infections, using sterile containers, transported in cold boxes, and processed on the same day.

2.2. Isolation and Identification of E. coli

Samples were enriched in buffered peptone water (BPW: 1.0% w/v) obtained by incubation at 37 °C for 18–24 h. After enrichment, one loopful of enriched broth was streaked onto Eosin methylene blue agar (EMBA) and EMBA with Cefotaxime (EMB + CX; for initial ESBL screening EMB + CX (1 μg mL⁻¹)). The inoculated plates were incubated at 37 °C for 24 hrs. After incubation, four small circular, purple-centered colonies with green metallic sheen morphology were selected. Presumptive isolates were subject to Gram stain and confirmed by biochemical tests using an E. coli identification kit (HiMedia; Hi E. coli TM Identification Kit, KB010). Biochemically confirmed E. coli isolates were further confirmed by PCR targeting the universal stress protein A (uspA) gene. The primers were (F: CGATACGCTGCCAATCAGT; R: ACGCAGACCGTAGGCCAGAT), producing an 884 bp product. The reaction was performed in a 25 μL total volume/tube mixture per previously described methods [31], Taq 2× Master Mix, Red 1.5 mM (AMPLIQON 5200300-1250, Massachusetts, USA). E. coli ATCC 25922 was used as a positive, Salmonella Typhimurium ATCC 14028 as a negative, and nuclease-free water as a no-template control. All confirmed E. coli isolates were stored at −20 °C until further use. All culture media, reagents, and antibiotics used in this study were from HiMedia, Maharashtra, India.

2.3. Phenotypic Characterization of Antimicrobial Resistance in E. coli Isolates

From each positive sample for E. coli (n = 291), a single isolate was randomly selected and subjected to antimicrobial susceptibility testing based on the standard disc diffusion assay as per the European Committee on Antimicrobial Susceptibility Testing [32]. Susceptibility was performed with gentamicin (GEN = 10 μg), amikacin (AMK = 30 μg), neomycin (NEO = 10 μg), ciprofloxacin (CIP = 5 μg), doxycycline (DOX = 30 μg), trimethoprim-sulfamethoxazole (COT = 25 (23:75/1:25 μg), chloramphenicol (CHL = 30 μg), ampicillin (AMP = 10 μg), amoxicillin + clavulanic acid (AMC = 20/10 μg), cefotaxime (CTX = 30 μg), ceftazidime (CAZ = 30 μg), and cefpodoxime (CPD = 10 μg). All E. coli isolates were classified as susceptible or resistant to the tested antimicrobial agents based on the measured zone diameters, in accordance with the breakpoints established by the EUCAST Clinical Breakpoints [32]. For quality control, E. coli ATCC 25922 was used each time for each batch of experiments Resistance to antimicrobials was defined as rare: <0.1%; very low: 0.1% to 1.0%; low: >1.0% to 10.0%; moderate: >10.0% to 20.0%; high: >20.0% to 50.0%; very high: >50.0% to 70.0% and extremely high: >70.0% [33].

2.4. Determination of Minimum Inhibitory Concentration (MIC) of Colistin in E. coli Isolates

The standard broth microdilution technique was used to assess the MIC of colistin sulfate (COL) among E. coli isolates (n = 291), employing cation-adjusted Mueller–Hinton broth (CAMHB, HiMedia) [34]. The test was performed in untreated polystyrene flat bottom 96-well plates. Two-fold serial dilutions of colistin sulfate, ranging from 0.25 to 128 mg/L, were prepared in CAMHB and inoculated with the isolates. The plates were incubated at 37 °C for 18 h. The endpoint was determined as the lowest concentration of colistin, which completely inhibited visible growth. Sterility control: CAMHB only, highest antibiotic control: CAMHB + 128 μg/mL of colistin, and lowest antibiotic control: CAMHB +0.25 μg/mL were maintained in duplicate in each plate. Escherichia coli (ATCC^®25922™) was used as a quality control for each batch of screening. The cut-off value was interpreted as per the EUCAST breakpoint (2 μg/mL for colistin sulfate).

2.5. Phenotypic Confirmation of Extended-Spectrum Beta-Lactamase (ESBL) in E. coli Isolates Using Double-Disk Diffusion Test

All E. coli isolates (n = 291) were tested for extended-spectrum beta-lactamase (ESBL) production using the double-disc diffusion test with commercial discs (HiMedia Hexa G-minus 24 catalog No. HX096) containing cefotaxime (CTX = 30 μg), cefotaxime + clavulanic acid (CEC = 30/10 μg), ceftazidime (CAZ = 30 μg), ceftazidime + clavulanic acid (CAC = 30/10 μg), cefpodoxime (CPD = 10 μg) and cefpodoxime + clavulanic acid (CCL = 10/5 μg) as described in previous studies [35]. The results were interpreted according to the Clinical and Laboratory Standards Institute [36]. Resistance to at least one of the three antibiotics (cefotaxime (≤27 mm), ceftazidime (≤22 mm), and cefpodoxime (≤17 mm) was considered positive in the screening test for possible ESBL production, and an increase in the zone diameter by ≥5 mm containing cephalosporin with clavulanic acid over the disks containing cephalosporin alone for any one of the groups indicated ESBL production. For quality control, E. coli ATCC 25922 was used each time for each batch of experiments.

2.6. Extraction of DNA from E. coli Isolates

Bacterial DNA was extracted using DNA purification kits (QIAamp DNA Micro Kit (50) Cat. No./ID: 56304) by QIAGEN N.V., Hulsterweg 82, 5912 PL Venlo, The Netherlands. DNA concentration and purity were measured using a Nanodrop spectrophotometer, NanoDrop® (Thermo Fisher Scientific Inc., 168 Third Avenue, Waltham, MA 02451, USA), with the A260/A280 and A260/A230 ratios recorded to assess purity. Extracted DNA was preserved at −20 °C until further use.

2.7. Determination of AMR Genes in E. coli Isolates

In this study, 35 AMR genes were investigated that belonged to seven antimicrobial groups: aminoglycosides {aac(3)-IIa (aacC2)a, aph(3)-IIa (aphA2)a, aac(6′)-Ib}, quinolones (qnrA, qnrB, OqxAB, qepA, qnrD, qnrS, aac(6′)-Ib-cr, qnrC), polymyxins (mcr-1, mcr-2, mcr-3, mcr-4, mcr-5), tetracycline (tetA, tetB, tetC, tetD, tetM), phenicols (catA1, catA2, cmlA1), sulfonamide (sul1 and sul2) and β-lactams {bla_TEM, bla_NDM, bla_OXA, bla_NDM, bla_CTX-M), bla_CTX-M group-1, bla_CTX-M group 2, bla_CTX-M group 9, bla_CTX-M group 8/25, (Genes, primers and PCR conditions described previously [37,38,39,40,41,42,43,44] (Supplementary Table S1). A 25 μL reaction mix consisted of 12.5 μL of Taq DNA Polymerase Master Mix RED (Amliqon^®), forward and reverse primers (1 μL of 10 pM for all the primers), 10.5 μL of nuclease-free water, and one μL of DNA template (~100 ng). Each PCR run included positive, negative, and no-template controls (NTC, nuclease-free water). Agarose gel (2.0%) electrophoresis separated the PCR products.

2.8. Statistical Analysis

Statistical analysis of this study was performed using R Studio (version 2024. 12. 0 + 467) (Copyright (C) 2024. The R Foundation for Statistical Computing, Posit Software, PBC) using the “aggregate”, “trend”, “tidyverse”, “gtsummary”, “AER”, “glm”, “ggplot”, “circlize”, “chordding”, “webshot”, and “devtools” “caret”, “irr”, “ggalluvial” packages.

For each sampling source and sampling point, antimicrobial resistance proportions were calculated by dividing the number of resistant E. coli isolates by the total number of isolates tested for that specific antimicrobial. For each proportion, exact binomial CIs with the Clopper–Pearson method were calculated. Similarly, for each sampling source and sampling point, ARG proportions were calculated by dividing the number of E. coli isolates that tested positive for a specific gene by the total number of isolates tested for that particular gene. The exact binomial 95% CIs with the Clopper–Pearson method were calculated for each proportion. Poisson regression was used to assess AMR differences across sampling days (days 1, 18, and 36).

Individual Poisson regression models were constructed for each antimicrobial. The number of E. coli tested in each sampling source represented the outcome variable. The predictor variables were represented by the source of sampling (BBF, hatchery, CBF, and RMS). The number of all E. coli tested in that source was included as the offset to account for the differences in the number of resistant isolates across the sources. Incidence rate ratios (IRRs), 95% CIs, and p-values were calculated for each regression model coefficient and were represented in graphs. A p-value ≤ 0.05 was considered significant. Compared to the referent category, an IRR of <1 indicated a decrease, and >1 indicated an increase in the rate of E. coli isolates in a given source.

Ward’s hierarchical single-linkage clustering method, using Euclidean distances, was employed to construct dendrograms for co-resistance patterns, the presence of ARGs, and clustering among E. coli isolates and across different sampling sources. Each dendrogram was illustrated in a heat map. The R software packages “heatmaps.2,”, “ggplot,” and the RColorBrewer library were used to create and visualize the dendrograms as heatmaps.

Pairwise and overall correlations among antimicrobial resistance were assessed using Pearson correlation coefficients. These correlations were visualized through chord diagrams, which depicted relationships between antimicrobials. The chord diagrams were created using the “chorddiag” and “devtools” R packages.

AMR phenotype–genotype concordance was analyzed by mapping phenotypic resistance to associated resistance genes in E. coli isolates. A predefined phenotype–genotype relationship was used to extract relevant genes for each resistant phenotype. Isolates exhibiting phenotypic resistance were identified, and the presence of corresponding resistance genes was quantified. The number of unique isolates carrying each resistance gene was determined, and the data were visualized using an alluvial plot to illustrate phenotype–genotype relationships. A custom color scheme was applied for better visualization. All analyses and visualizations were performed in R using the “ggplot2” and “ggalluvial” packages.

3. Results

3.1. Phenotypic Antimicrobial Resistance

Out of 332 samples, E. coli was isolated from 291 (88%) of the samples.

Table 2. Prevalence of antimicrobial resistance in E. coli isolates of the broiler production chain.

Antimicrobial		Proportion of AMR in the Broiler Production Chain						Proportion of AMR in the Commercial Broiler Farm Crop Cycle
Class	Drug A	BBF (N = 38) B	Hatchery (N = 41) B	CBF (N = 117) B	RMS (N = 95) B	Total (N = 291)	95% CI C	Day 1 (N = 36) B	Day 18 (N = 36) B	Day 36 (N = 45) B	Total (N = 117) B	95% CI C
		n (%) D	n (%) D	n (%) D	n (%) D	n (%) D		n (%) D	n (%) D	n (%) D	n (%) D
Aminoglycosides	GEN	21 (55.26)	13 (31.71)	70 (59.83)	59 (62.11)	163 (56.01)	50.1–61.8	23 (63.89)	18 (50.00)	29 (64.44)	70 (59.83)	50.36–68.78
	AMK	12 (31.58)	10 (24.39)	37 (31.62)	42 (44.21)	101 (34.71)	29.25–34.71	10 (27.78)	14 (38.89)	13 (28.89)	37 (31.62)	23.34–40.87
	NEO	10 (26.31)	6 (14.63)	37 (31.62)	13 (13.68)	66 (22.68)	18–27.93	14 (38.89)	9 (25.00)	13 (28.89)	36 (30.77)	22.57–39.97
Fluoroquinolones	CIP	28 (73.68)	29 (70.73)	100 (85.47)	74 (77.89)	231 (79.73)	74.64–84.19	32 (88.89)	32 (88.89)	36 (80.00)	100 (85.47)	77.76–91.3
Tetracycline	DOX	35 (92.11)	41 (100.00)	114 (97.44)	92 (96.84)	282 (96.91)	94.21–98.58	36 (100.00)	35 (97.22)	43 (95.56)	114 (97.44)	92.69–99.47
Folate pathway antagonists	COT	17 (44.74)	18 (43.9)	57 (48.72)	46 (48.42)	138 (47.72)	41.57–53.33	19 (52.78)	15 (41.67)	23 (51.11)	57 (48.72)	39.37–58.13
Phenicols	CHL	10 (26.32)	8 (19.51)	34 (29.06)	15 (15.79)	67 (23.02)	18.31–28–29	10 (27.78)	9 (25.00)	15 (33.33)	34 (29.06)	21.04–38.17
Polymyxins	COL	3 (7.89)	8 (19.51)	31 (26.50)	18 (18.95)	60 (20.61)	16.12–27.73	1(2.78)	15 (41.67)	15 (41.67)	31 (26.5)	18.77–35.45
Penicillin	AMP	20 (52.63)	25 (60.98)	88 (75.21)	68 (71.58)	201 (69.07)	63.41–74.34	30 (83.33)	28 (77.78)	30 (66.67)	88 (75.21)	66.36–82.73
β-Lactamase combination agents	AMC	22 (57.89)	19 (46.34)	41 (35.04)	36 (37.89)	118 (40.55)	34.86–46.44	9 (25.00)	11 (30.56)	20 (44.44)	40 (34.19)	25.67–43.53
Third-generation cephalosporins (3CGR)	CTX	19 (50.00)	21 (51.22)	63 (53.85)	61 (64.21)	164 (56.36)	50.45–62.14	19 (52.78)	16 (44.44)	28 (62.22)	63 (53.85)	44.39–63.1
	CAZ	15 (39.47)	22 (53.66)	57 (48.71)	63 (66.32)	157 (53.95)	48.04–59.78	16 (44.44)	11 (30.56)	29 (64.44)	56 (47.86)	38.54–57.29
	CPD	33 (86.84)	39 (95.12)	99 (84.62)	79 (83.16)	250 (85.91)	81.38–89.7	32 (88.89)	30 (83.33)	37 (82.22)	99 (84.62)	76.78–90.62
3CGR average	-	58.77	(66.66)	(62.40)	(71.23)	(65.41)	62.25–68.56	(62.03)	(52.77)	(69.63)	(62.11)	53.24–70.34
ESBL E Producer		17(44.74)	18(43.90)	40 (34.19)	46(48.42)	121 (41.58)	35.86–47.48	14 (38.89)	17 (47.22)	17(37.78)	48 (41.03)	32.02–50.5
MDR F		37 (97.37)	40 (97.56)	113 (96.58)	92 (96.84)	282 (96.84)	94.21–98.58	35 (97.22)	35 (97.22)	43 (95.56)	113 (97.44)	91.48–99.06

^A Gentamicin (GEN), amikacin (AMK), neomycin (NEO), ciprofloxacin (CIP), doxycycline (DOX), trimethoprim-sulfamethoxazole (COT), chloramphenicol (CHL), ampicillin (AMP) and amoxicillin and clavulanic acid (AMC), cefotaxime (CTX), ceftazidime (CAZ), and cefpodoxime (CPD). ^B N= Total number of tested isolates, ^C CI: Exact binomial 95% confidence interval. ^D n = Total number of positive isolates. ^E Extended—spectrum β—lactamase producers, ^F MDR: Multi-Drug Resistance.

describes the resistance to 13 antimicrobials in E. coli isolates across the broiler production chain. In the present study, 100% of the isolates obtained were resistant to at least one of the antimicrobials tested. Overall, 282/291 (97%) were multidrug-resistant as defined by [13]. Among the complete broiler production chain isolates, irrespective of the sample, extremely high resistance (>70%) was observed to doxycycline, ciprofloxacin, very high resistance (>50% to 70%) to gentamicin, and high resistance (>20% to 50%) to trimethoprim-sulfamethoxazole, amikacin, chloramphenicol, neomycin, and colistin (Table 1). Antimicrobial susceptibility was tested at different time points during broiler crop cycles (CBF), revealing extremely high to high resistance on day one, except for colistin, which was moderate. At the midpoint of the crop cycle (day 18) and at the end of the crop cycle (day 36), all isolates exhibited extremely high to high resistance (Table 2).

Extremely high resistance was detected in E. coli isolates from the farm’s external environmental samples of BBFs to doxycycline, ciprofloxacin, and amikacin, as well as very high resistance to neomycin, gentamicin, trimethoprim-sulfamethoxazole, and chloramphenicol, and high resistance to colistin. Among the CBF isolates from farm external environmental samples, extremely high resistance was detected to doxycycline and ciprofloxacin, with very high resistance to gentamicin and trimethoprim-sulfamethoxazole, and high resistance to all other antimicrobials tested (Table 2). The Poisson regression model showed no significant difference among different days of sample points for any of the tested antimicrobials (Figure 2).

3.1.1. Colistin Resistance

In the complete broiler production chain, 21% of the isolates were resistant (MIC > 2 µg/mL) to colistin. A significant difference (p < 0.0001) was observed in the number of colistin-resistant isolates obtained from BBFs (7.89%), hatcheries (19.51%), CBFs (26.50%), and RMSs (18.95%). There was an increasing trend for resistance to colistin in E. coli isolated from CBFs at different time points during the crop cycle (p < 0.001). Colistin resistance was present in 2.78% of isolates on day 1, increasing to 41.67% on days 18 and 36.

On average, 65% of isolates were third-generation cephalosporin-resistant (3GCR), with a higher resistance rate in the final stage of the broiler production chain (RMS; 71%) and a lower rate at the beginning of the broiler breeding farm (BBF; 58%) (Table 2).

3.1.2. Extended-Spectrum Beta-Lactamase (ESBL)-Producing E. coli

In the present study, 121 out of 291 isolates (41.58%) were ESBL irrespective of the sampling source. The presence of ESBL-producing E. coli was highest in RMSs (48.42%), followed by BBFs (45.0%), hatcheries (44%), and lowest at CBFs (34%). Concerning the E. coli isolated from the CBF crop cycle, the highest number of ESBL producers was detected on day 18 of the placement of chicken in the farm (47%), followed by day one (39%), and the lowest on day 36 (38%), with no significant difference (p < 0.659) among the days of the crop cycle (

Table 3. Prevalence of MDR-ESBL-producing, fluoroquinolone-ESBL-producing, and colistin-ESBL-producing E. coli in the broiler production chain.

Sampling Source			MDR-ESBL-Producer	Fluoroquinolones-ESBL-Producer	Colistin-ESBL-Producer
Proportion of AMR in the broiler production chain	BBFs (N A = 13)	n B (%)	13 (100)	7 (53.85)	1 (7.69)
	Hatchery (N = 11)	n (%)	11 (100)	7 (63.64)	3 (27.27)
	CBFs (N = 49)	n (%)	49 (100)	46 (93.38)	16 (32.65)
	RMSs (N = 48)	n (%)	46 (95.85)	37 (77.08)	10 (20.83)
	Total (N = 121)	n (%)	119 (98.35)	97 (80.17)	30 (24.79)
	95% CI C		64.16–99.8	71.94–86.86	17.4–33.46
Proportion of AMR in the commercial broiler farm crop cycle	Day 1 (N = 14)	n (%)	14 (100)	14 (100)	1 (7.14)
	Day 18 (N = 12)	n (%)	12 (100)	12 (100)	4 (33.33)
	Day 36 (N = 23)	n (%)	23 (100)	20 (86.96)	11 (47.83)
	Total	n (%)	49 (100)	46 (93.38)	16 (32.65)
	95% CI C		92.75–100	83.13–98.72	19.95–47.54

^A N = The total number of ESBL isolates tested, ^B n = Total number of positive isolates. ^C CI: Exact binomial 95% confidence interval.

). Among the ESBL isolates, 119/121 (98%) were MDR-ESBL producers. The presence of MDR-ESBL-producing E. coli was 100% at BBFs, hatcheries, and CBFs and 96% in RMSs. In the CBF crop cycle, all the isolates were MDR-EBL producers in all three time points. Among the ESBL E. coli isolates, 96/121 (80%) were fluoroquinolone-resistant ESBL producers, and 30/ 121 (25%) were colistin resistant (Table 3).

3.1.3. Clustering Dendrogram of E. coli Isolates

The clustering dendrogram (heatmap) for AMR in E. coli isolates is presented in Figure 1. The red color indicates resistance, and the blue color indicates susceptible isolates. Figure 1 highlights patterns of resistance within sources and antimicrobials. The clustering demonstrates the distribution of antimicrobial resistance across different sampling sources (x-axis) and antimicrobial agents tested (y-axis). The heatmap revealed distinct patterns of AMR across the broiler production chain. Resistance rates varied by antimicrobials, with the highest resistance to doxycycline, cefpodoxime, and ciprofloxacin, and the lowest to neomycin and colistin. The CBF isolates showed higher resistance among the sources, followed by RMSs, BBFs, and the lowest at hatcheries.

3.1.4. Assessing Pairwise and Total Correlations Among AMR of E. coli Isolates

Figure 2 illustrates the total and pairwise correlations of resistance to different antimicrobial agents in E. coli isolates of the complete broiler production chain. Among E. coli isolates, the highest total correlation of 2.3 was observed for amikacin, followed by 2.2 for ciprofloxacin, cefotaxime, and ceftazidime, 2.1 for neomycin, ampicillin, 2 for gentamycin, 1.9 for chloramphenicol and trimethoprim-sulfamethoxazole, 1.8 for doxycycline, 1.7 for amoxicillin + clavulanic, and 1.4 for cefpodoxime. For pairwise correlations, the highest was observed between cefotaxime and ceftazidime (0.7) and ampicillin and ceftazidime (0.4).

3.1.5. Regression Analysis

Figure 3 shows the results of the Poisson regression models for each antimicrobial, assessing the associations between the rates of AMR at different sampling sources (BBFs, hatcheries, CBFs, RMSs). In this model, compared to BFF, no antimicrobials showed a significant increase in various segments of the broiler production chain, except for colistin, which showed a significant increase in CBFs compared to the BBFs, hatcheries, and RMSs. For colistin, the resistance rate was 2.47 for hatcheries, 3.36 for CBFs, and 2.40 for RMSs compared to BBFs (Figure 3).

3.2. Prevalence of ARGs in E. coli Isolates in the Broiler Production Chain

Table 4. Prevalence of ARGs in E. coli isolates.

Class of Antimicrobials	Target Genes A	Proportion of ARGs in the Broiler Production Chain						Proportion of ARGs in the Broiler Crop Cycle
		BBF (N = 38)A	Hatchery (N = 41) A	CBF (N = 117) A	RMS (N = 95) A	Total (N = 291) A	95% CI C	Day 1 (N = 36) A	Day 18 (N = 36) A	Day 36 (N = 45) A	Total (N = 117) A	95% CI C
		n (%) B	n (%) B	n (%) B	n (%) B	n (%) B		n (%) B	n (%) B	n (%) B	n (%) B
β Lactams	blaTEM	5 (13.16)	20 (48.78)	59 (50.43)	36 (37.89)	120 (41.24)	35.52–47.13	22 (61.11)	21 (58.33)	16 (55.56)	59 (50.43)	41.03–59.8
	blaSHV	0(0)	0(0)	8 (6.84)	13 (13.68)	21 (7.22)	4.52–10.82	1 (2.78)	0 (0)	7 (15.56)	8 (6.84)	3–13.03
	blaOXA	0(0)	1 (2.44)	4 (3.42)	5 (5.26)	10 (3.44)	1.66–6.23	1 (2.78)	1 (2.78)	2 (4.44)	4 (3.42)	0.94–8.52
	blaCTX-M	5 (13.16)	13 (31.71)	32 (27.08)	32 (32.63)	82 (28.18)	23.08–33.72	10 (27.8)	8 (22.22)	14 (31.11)	32 (27.35)	19.52–36.36
	blaCTX-M Group 1	5 (13.16)	9 (21.95)	27 (23.08)	31 (32.63)	72 (24.74)	19.89–30.11	7 (19.44)	8 (22.22)	12 (26.67)	27 (23.08)	15.79–31.77
	blaCTX-M Group 2	0(0)	2 (4.88)	2 (1.71)	2 (2.11)	6 (2.06)	0.76–4.43	1 (2.78)	0 (0)	1 (22.22)	2 (1.71)	0.21–6.04
	blaCTX-M Group 9	0(0)	2 (4.88)	1 (0.85)	1 (1.05)	4 (1.37)	0.38–3.48	0 (0)	0 (0)	1 (22.22)	1 (0.85)	0.02–4.67
	blaCTX-M Group 8/25	0(0)	2 (4.88)	1 (0.85)	1 (1.05)	4 (1.37)	0.38–3.48	1 (2.78)	0 (0)	0 (00)	1 (0.85)	0.02–4.67
Quinolone	qnrB	12 (31.58)	22 (53.66)	71 (60.68)	83 (87.37)	188 (64.6)	58.81–70.1	13 (36.11)	28 (77.78)	30 (66.67)	71 (60.68)	51.27–69.59
	qepA	9 (23.68)	15 (36.59)	25 (21.37)	7 (7.37)	56 (19.24)	14.88–24.25	14 (38.89)	4 (11.11)	7 (15.56)	25 (21.37)	14.33–29.91
	qnrD	0(0)	0(0)	14 (11.97)	4 (4.21)	18 (6.19)	3.71–9.6	2 (5.55)	1 (2.780	11 (24.44)	14 (11.97)	6.7–19.26
	qnrS	12 (31.58)	16 (39.02)	74 (63.25)	52 (54.74)	154 (52.92)	47.01–58.77	16 (44.44)	22 (61.11)	36 (80.00)	74 (63.25)	53.84–71.97
	aac(6′)-Ib-cr	11 (28.95)	8 (19.51)	35 (29.91)	35 (36.84)	89 (30.58)	25.34–36.23	9 (25)	6 (16.67)	20 (44.44)	35 (29.91)	21.8–39.07
Polymyxins	mcr-1	0(0)	15 (36.59)	42 (35.9)	25 (26.32)	82 (28.18)	23.08–33.72	5 (13.89)	21 (58.33)	16 (35.56)	42 (35.9)	27.24–45.29
	mcr-2	3 (7.89)	3 (7.32)	29 (24.79)	30 (31.58)	65 (22.34)	17.68–27.56	9 (25)	9 (25)	11 (24.44)	29 (24.79)	17.27–33.62
	mcr-3	2 (5.26)	4 (9.76)	25 (21.37)	33 (34.74)	64 (21.99)	17.37–27.2	9 (25)	9 (25)	7 (15.56)	25 (21.37)	14.33–29.91
	mcr-4	1 (2.63)	0(0)	24 (20.51)	28 (29.47)	53 (18.21)	13.95–23.14	9 (25)	9 (25)	6 (13.33)	24 (20.51)	13.61–28.97
	mcr-5	0 (0)	3 (7.32)	27 (23.08)	27 (28.42)	57 (19.59)	15.19–24.62	8 (22.22)	12 (33.33)	7 (15.56)	27 (23.08)	15.79–31.77
Tetracycline	tetA	10 (26.32)	32 (78.05)	88 (75.21)	72 (75.79)	202 (69.42)	63.77–74.66	23 (63.89)	28 (77.7)	37 (82.22)	88 (75.21)	66.38–82.73
Phenicol	catA1	0 (0)	8 (19.51)	24 (20.51)	6 (6.32)	38 (13.06)	9.41–17.48	2 (5.55)	7 (9.44)	15 (33.33)	24 (20.51)	13.61–28.97
	catA2	0 (0)	7 (17.07)	11 (9.4)	3 (3.16)	21 (7.22)	4.52–10.82	2 (5.55)	5 (13.89)	4 (8.89)	11 (9.4)	4.79–16.2
	cmlA1	3 (7.89)	12 (29.27)	20 (17.09)	12 (12.63)	47 (16.15)	12.12–20.89	10 (27.78)	5 (13.89)	5 (11.11)	20 (17.09)	10.77–25.16
Folate pathway antagonists	sul1	2 (5.26)	4 (9.76)	31 (26.5)	9 (9.47)	46 (15.81)	11.81–20.52	5 (13.89)	8 (22.22)	18 (40.00)	31 (26.5)	18.77–35.45
	sul2	1 (2.63)	2 (4.88)	7 (5.98)	0(0)	10 (3.44)	1.66–6.23	0 (0)	2 (5.55)	5 (11.11)	7 (5.98)	2.44–11.94
Total ARGs D		14/35 (40.00)	21/35 (60.00)	24/35 (68.57)	23/35 (65.71)	24/35 (68.57)	50.71–83.15	22/35 (62.86)	20/35 (57.14)	23/35 (65.71)	24/35 (68.57)	50.71–83.15

^A N = Total number of tested isolates, ^B n = Total number of positive isolates. ^C CI: Exact binomial 95% confidence interval. ^D ARGs: Antimicrobial-resistant genes.

describes the presence of 35 ARGs among E. coli isolates across the broiler production chain. In the complete broiler production chain, regardless of the sampling source, 24 out of 35 tested ARGs were detected (67%). Among β-lactam genes, bla_TEM was present in 41% of isolates, followed by bla_CTX-M 28%, among which bla_CTX-M group 1 was 25%, bla_SHV 7%, bla_OXA 3% and bla_CTX-M group 9 (1%). Among quinolone-resistant genes, the prevalence of qnrB was the highest at 65%, followed by qnrS at 53%, and lowest for qnrD at 6%. However, qnrA, OqxAB, and qnrC were not present. Among the five tetracycline resistance genes, only tetA was detected in 69% isolates. The phenicol resistance genes cmlA1 and catA1 were detected in 16% and 13% of isolates, and the sulfonamide resistance genes sul1 and sul2 were detected in 16% and 3%, respectively. The prevalence of mcr genes in E. coli isolates was 22%, with mcr-1 being the most commonly identified gene, followed by mcr-2, 3, 5, and 4 (p < 0.01). The presence of mcr genes varied across different sources, with mcr-2, 3, and mcr-4 detected in BBFs isolates, while all genes except mcr-4 were found in hatcheries. All five mcr genes were present in CBFs and RMSs isolates and E. coli isolates from the broiler crop cycle at all three time points. All the ARGs from BBFs to RMSs were significantly increased except for bla_CTX-M Group 2, bla_CTX-M Group 9, and bla_CTX-M Group 8/25 (Table 4, Figure 4).

3.3. Prevalence of ARGs in E. coli Isolates in the Broiler Crop Cycle

Among broiler crop cycle β-lactam genes, bla_TEM was prevalent in 50% of isolates, followed by bla_CTX-M in 27%, and none of the bla_NDM genes were detected. For quinolone genes, the presence of qnrS was highest at 63% and qnrD was lowest at 12%. However, qnrA, OqxAB, and qnrC were not present. Among the five tetracycline resistance genes, only tetA was detected in 75% isolates. The phenicol resistance genes cmlA1 (27%), catA1 (21%), and catA2 (7%), as well as the sulfonamide resistance genes sul1 and sul2, were detected in 16% and 9% of the samples, respectively. All five mcr genes were detected in the broiler crop cycle, with mcr-1 being the most commonly identified gene, followed by mcr-2, 5, 3, and 4 (p < 0.01) (Table 4, Figure 4).

Antimicrobial Resistance Genes (ARGs) in E. coli Isolated from Different Sample Types

ARGs were detected among E. coli isolated from multiple samples across different segments of the broiler production (BBFs, hatcheries, CBFs, and RMSs). Among the broiler breeder’s samples, different ARGs were detected in the external environment, cloacal swabs, egg yolk, feed bags, and feeders. Various AMR genes were detected in E. coli isolates from the different sample types from the BBFs (

Table 5. Proportions of ARGs in E. coli isolated from different types of samples.

No.	Type of Samples	AMR Genes Detected (%)
BBFs
1	External environment	qnrB, aac(6′)-Ib-cr, qnrS, and tetA (33)
2	Cloacal swabs	mcr-2 & cmlA1 (29), qnrB, aac(6′)-Ib-cr, qnrS & (14),
3	Egg yolk	blaTEM, qnrB & tetA (14).
4	Feed from feed bags	qnrS and aac(6′)-Ib-cr (14).
5	Feed from feeders	qnrB, qepA, aac(6′)-Ib-cr, qnrS, tetA, cmlA1 and sul1 (14)
Hatcheries
1	Hatcheries internal environments	tetA (100), blaTEM, qnrS, qepA (67), qnrB, aac(6′)-Ib-cr, mcr-2, and cmlA1 (33),
2	Chick trays	tetA (100), qepA, qnrS, aac(6′)-Ib-cr (66.67), blaTEM, blaCTX-M, blaCTX-M group1, qnrB, mcr-1, cmlA2 and sul2 (33)
3	Hatchers samples	tetA (100), qnrB and sul1 (67), blaTEM, blaCTX-M, qnrS, aac(6′)-Ib-cr, mcr-1, mcr-2 and cmlA1(33)
4	Incubator air tunnels	tetA, qnrB and qnrS (67), blaTEM, blaCTX-M, group, qepA, aac(6′)-Ib-cr and cmlA1 (33)
5	Incubators	blaTEM, qnrB, tetA, and mcr-1 (33).
6	Meconium dropping	tetA (100), qnrB & qepA (67) and blaTEM, qnrS, cmlA1 & mcr-1 (33)
7	Eggs setting room	mcr-5 & catA1 1/3(33)
8	In the workers’ hand swabs	qnrS and tetA (60), qnrB and qepA mcr-3 catA1, cmlA1 (40), blaTEM, aac(6′)-Ib-cr and mcr-1 (20).
CBFs
1	Farm internal environment samples (Boot socks inside the farm)	tetA (80), qnrS (67), qnrB (60), mcr-1 (47), blaTEM, qepA, mcr-2,(27), catA1 (33), aac(6′)-Ib-cr, mcr2-5, and, sul1 (20), catA2, cmlA1 (13)
2	farm external environment (Boot socks outside the farm)	tetA (67), blaTEM, qnrS (56), aac(6′)-Ib-cr 4/9(44), qepA, mcr-2-5 and catA1 (33), qnrD and, cmlA1 (22), mcr-1, sul2 (11)
3	Fecal swabs	tetA (87), qnrS (73), blaTEM (60), qnrB (53), aac(6′)-Ib-cr (47), sul1, mcr-1 mcr-2, and mcr-5 and cmalA1 (27), mcr-3 and mcr-4 (20), sul2 and blaCTX-M (13), blaSHV, bla group 1, qepA and catA1 (7)
4	Feed samples from feeders	tetA (80), qnrB & qnrS (67), blaTEM (53), catA1 6/15 (40), mcr-1 (33), sul1 and aac(6′)-Ib-cr (27), qepA, mcr-2 mcr-3, mcr-4 and mcr-5 and cmalA1 (20) and blaCTX-M (Universal & bla group (7)
5	Feed from feed bags	blaCTX-M, cmlA1, mcr-2 and mcr-3 and mcr-4 (7), blaTEM, mcr-5, sul1 and sul2 (13), qnrS, qepA mcr-1 and tetA (27)
6	Water nipple samples	qnrS (47) tetA (40), blaTEM (13), mcr-2 (33), qnrB, mcr-3-5 (20), blaSHV, blaCTX-M and bal group1, catA2, sul1 and aac(6′)-Ib-cr \(13) and blaOXA (7)
RMSs
1	Meat rinsing water	qnrB, qnrS, mcr-3, catA1 and tetA (67), blaTEM, aac(6′)-Ib-cr, catA1, mcr-2 mcr-4 and mcr-5 (33)
2	Ileal contents	tetA (100), qnrB (93%), qnrS (73), & blaTEM (40), mcr2-4 (33), mcr-5 (27), mcr-1 (20), qepA, catA1, cmlA1 (13), blaCTX-M blaSHV, bla group 1 and sul1 1/15 (67)
3	Caecal contents	tetA (100), qnrB (93.33), mcr-1 (53), cmlA1 (47), blaTEM, aac(6′)-Ib-cr & qnrS (33), mcr2-4 (27), mcr-5 (20), blaCTX-M, blaSHV, qepA and sul1 (13), blaOXA, catA1 and blaTEM group1 (7)
4	chicken carcass	qnrD (80%), tetA (73), qnrS (67), blaTEM (47), aac(6′)-Ib-cr 5/20(33), blaSHV, mcr1-4 (27), mcr-5, cmlA1 and sul2(20), qepA and catA1 (13), blaCTX-M, group1, catA2 group8/25 (7)
5	Cutting knife	qnrB & tetA 2/3 (67), blaTEM, qepA, qnrS, mcr-3, 1/mcr-5 (33),
6	Cutting board	blaCTX-M & bla group1, bla group2, qnrB, qnrS, aac(6′)-Ib-cr, mcr-2 and mcr-3 (33)

). The external environment had qnrB, aac(6′)-Ib-cr, qnrS, and tetA (33%), while cloacal swabs carried mcr-2, cmlA1, and (29%), egg yolk, feed from bags, and feeders harbored multiple resistance genes, including bla_TEM, qnrB, tetA, qepA, and sul1. AMR genes were widely detected across hatcheries. tetA was detected in all hatchery internal environments, including the chick tray, hatcher, and meconium samples (100%). Other genes, including bla_TEM, qnrS, qnrB, qepA, aac(6′)-Ib-cr, mcr-1, mcr-2, cmlA1, and sul1, were identified at varying frequencies (33–67%) (Table 5). Workers’ hand swabs carried multiple resistance genes, with qnrS and tetA being the most common (60%). In CBF samples, ARGs were detected in the E. coli isolates from six different types of samples: farm internal and external environment samples, feed samples from feed bags, feeders, fecal swabs, and drinkers. AMR genes were commonly distributed across different sample types in the CBFs, indicating potential environmental and feed-related sources of AMR. tetA was the most prevalent, found in farm environments, fecal swabs, and feed samples (67–87%). Other common genes included qnrS, qnrB, bla_TEM, mcr-1, qepA, sul1, and aac(6′)-Ib-cr at varying levels (7–67%)(Table 5). Water and feed samples harbored multiple resistance genes, including bla_CTX-M, bla_SHV, cmlA1, mcr-2-5, and catA1, though at lower frequencies (7–40%). In RMSs, ARGs were detected in E. coli from six different types of samples, including cutting boards, chicken carcasses, ileal and caecal contents, cutting knives, and meat rinsing water. In RMSs, tetA was present in ileal and caecal contents (100%) and in meat rinsing water (67%). qnrB, qnrS, bla_TEM, mcr-1-5, and aac(6′)-Ib-cr were found at varying frequencies (7–93%). Concerningly, E. coli from chicken carcasses, cutting knives, and cutting boards carried multiple resistance genes, including bla_CTX-M, bla_SHV, qepA, cmlA1, and sul1 (7–33%) (Table 5).

3.4. Correlation Between AMR Phenotype and Genotype

A Sankey diagram (Figure 5) visually represents the correlation between AMR phenotypes and genotypes, illustrating the association of resistance characteristics by linking phenotypic resistance patterns (e.g., tetracycline or quinolone resistance) to their corresponding ARGs (e.g., tetA or qnrB). The thickness of the connections represents the strength or frequency of these associations, making it easier to identify dominant genotypic markers contributing to phenotypic resistance. This visualization facilitates a deeper understanding of how resistance traits are distributed and interlinked, aiding in the assessment of potential intervention points within the broiler production chain. The highest correlation was between doxycycline resistance and the tetA gene, with almost total correlation, followed by CIP, which had the highest correlation with qnrB and qnrS. Similarly, AMP, AMC, CPD, CTX, and CAZ showed a higher correlation with bla_TEM and bla_CTX-M and a low correlation with bla_SHV and bla_OXA, and COL with all five mcr genes, and CHL showed the highest correlation with cmlA1, followed by catA1 and catA2.

4. Discussion

This study investigated the phenotypic and genotypic resistance determinants in E. coli isolates across the broiler production chain (broiler breeder farms, hatcheries, commercial broiler farms, and retail meat shops) in Bengaluru City and the Kolar district of Karnataka, India. Overall, E. coli isolates exhibited extremely high to high resistance to the majority of antimicrobials. In this study, 97% of the isolates were MDR, 42% were ESBL producers, 65% were 3CGR, and 21% were colistin-resistant. Among the 35 ARGs tested, 24 (68.57%) were identified in isolates. The prevalence of ARGs among E. coli was high across the broiler production chain, increasing from BBFs to hatcheries and CBFs.

This study detected extremely high to high AMR in all antimicrobial classes. MDR across the broiler production chain and crop cycles, revealing consistently high resistance regardless of the sampling source. There are limited data available on the AMR phenotype across the entire broiler production chain; however, the AMR detected in this study was higher than the previous studies in broiler breeder farms (3–57%) in Canada [8], and hatcheries in Europe (30%) [45]. Similarly, high AMR (5–90%) and MDR (89%) were reported in E. coli from CBFs in India [16,46] and RMSs (45–100%) in Bangladesh [47]. The detection of highly resistant E. coli in the entire broiler production chain in this study suggests that the broiler production chain and its environment, feed, water, and their farm and processing facilities could maintain and spread resistant E. coli strains [48], which could be transmitted to humans [49,50].

This study also highlighted that BBFs and hatcheries, as the earliest stages of the production chain, may serve as reservoirs for antimicrobial-resistant E. coli and might transmit these bacteria downstream, posing a risk to CBFs through vertical transmission via eggs [7].

Analysis of the various sample sources indicated that AMR was distributed across all sample sources in the production chain and its environment. This study indicated that AMR was present throughout the broiler crop cycle (Days 1, 18, and 36); however, the resistance to several drugs detected on day one is essential, indicating that in addition to AMU in broiler farms, other factors, such as the transmission of antimicrobial resistant bacteria from the environment, and vertical transmission of AMR from breeders to eggs to hatcheries and subsequently to CBFs [51], may be responsible for this finding.

In this study, several samples, including chicken carcass samples obtained from retail outlets and meat rinsing water used for washing the chicken carcass, revealed high AMR to several antimicrobials, supporting the idea that chicken carcass processing environments are a significant source of resistant E. coli and ARGs [21,47,52]. This type of retail chicken shop receives chicken daily from different suppliers and processes it in a small area. Slaughtering, dressing, cutting, and packing are done in the exact location, which increases the chances of cross-contaminating carcasses with AMR bacteria and subsequently transferring these resistant bacteria to the consumers.

The presence of MDR and ESBL-producing E. coli isolates poses a risk to poultry, humans, and the environment [53,54]. The high resistance in RMSs could be due to poor processing and hygienic practices. The absence of a cooling system in this type of outlet may further allow the bacteria to replicate and facilitate cross-contamination by staff or equipment.

The increased resistance to quinolones and the high prevalence of quinolone resistance genes throughout the entire broiler production chain underscore the need for effective stewardship programs. This may be due to their current use to treat salmonellosis and colibacillosis with fluoroquinolones, as resistance to other antimicrobials (e.g., ampicillin, trimethoprim-sulfamethoxazole, and chloramphenicol) has been increasing for several years. [55]. Tetracycline resistance development is possibly associated with the long-term and widespread use of this agent. Although trimethoprim and sulfamethoxazole are widely used in broilers for bacterial infection, the prevalence of sul1 and sul2 in our study was low in E. coli isolates. However, this study detected a high level of phenotypic resistance to this drug in the broiler production chain, possibly due to other genes or mechanisms that impact the selection of resistance to this class of antimicrobials. This finding is in agreement with previous studies that reported the presence of several genes for folate pathway antagonists [56,57,58,59].

Colistin is the last-resort antimicrobial for MDR pathogens in humans [22], and the use of colistin in animal feed has been banned in several countries following the recommendations of the World Health Organization [60,61]. Although the use of colistin in poultry is banned in India [26], this study found that 21% of E. coli isolates were phenotypically resistant to colistin. Resistance to colistin has been linked to selective pressure from decades of bacterial exposure to this antibiotic [62]. A ban on colistin use is expected to remove the selective pressure and might reduce colistin resistance in the long term. However, the restriction on colistin use in animal production alone may not be an effective means of mitigating resistance. The resistance genes located on plasmids with other antimicrobial resistance determinants may promote co-selection, especially if those drugs remain in use. Notably, mcr-1 confers cross-resistance to bacitracin, a widely used antibiotic in food animals, raising concerns that the use of bacitracin could indirectly influence colistin resistance [63].

The widespread occurrence of fluoroquinolone-resistant ESBL-producing E. coli (80%) further exacerbates the AMR burden [64]. The highest prevalence in CBFs (93.38%) suggests that such bacteria may be amplified within the farm environment, potentially through selective pressure from antimicrobial use or horizontal gene transfer. The observation that the lowest number of fluoroquinolone-resistant ESBL producers was found on day 36 (86.96%), while days 1 and 18 showed a 100% prevalence, indicates that resistance dynamics may fluctuate throughout the production cycle.

The detection of MDR-ESBL, fluoroquinolone-resistant, and colistin-resistant E. coli across different production stages underscores the urgent need for stringent antimicrobial stewardship, improved biosecurity measures, and continuous surveillance to mitigate the spread of highly resistant strains within the poultry industry and beyond.

The pairwise and overall correlations among antimicrobial resistance patterns in E. coli isolates provided a representation of co-resistance relationships between different antimicrobial agents. These correlations were assessed using Pearson correlation coefficients and visualized through chord diagrams, which revealed clusters of antibiotics that frequently share resistance mechanisms. Such co-resistance patterns provide essential insights into potential genetic or biochemical pathways underlying antimicrobial resistance and can inform empirical treatment strategies. Moreover, understanding these relationships is crucial for antimicrobial stewardship, as it enables the optimal use of antibiotics.

The presence of ARGs in E. coli from hatcheries may be due to the horizontal spread of resistant bacteria rather than the acquisition of resistance [65]. The presence of ARGs in E. coli isolated from egg trays, incubator air tunnels, and yolk sacs from dead chicks suggests that these genes are prevalent throughout hatcheries and that there is always the possibility of transferring them to other bacterial species. Therefore, controlling E. coli in hatcheries is essential in the poultry supply chain.

In the RMSs, resistance genes were frequently identified in meat processing environments, particularly in ileal and cecal contents, meat rinsing water, and processing surfaces. Additionally, the presence of several mcr, bla_CTX-M, bla_SHV, and quinolone-resistant genes on cutting boards and knives highlights the risk of transmission of these genes to consumers. Overall, the widespread distribution of AMR genes in E. coli across various poultry production stages highlights the need for stricter biosecurity measures, enhanced hygiene practices, and responsible antimicrobial use to mitigate the persistence and spread of resistance [64,66]. The lack of studies analyzing the presence of ARGs at the sample level indicates a need for further research. This will help researchers focus on the sampling method and policymakers concentrate on managing AMR more specifically.

Out of five, three mcr genes (mcr-2, 3, and 4) were detected in E. coli isolates from BBFs, and four mcr genes (mcr-1, 2, 3, and 4) were detected in E. coli isolates from hatcheries. To the best of our knowledge, this is the first report on the detection of mcr genes from hatcheries, and it might be due to the vertical transmission of these genes from BBFs [7]. The presence of mcr genes in hatcheries is also of concern, as it may pose a high risk to the subsequent poultry chain as well as humans due to the horizontal transfer of resistance genes [67].

The detection of a high number of mcr genes in E. coli isolates from the broiler production cycle is a significant finding. E. coli can exist as both a commensal and a pathogenic organism in animals, humans, and the environment. The presence of mcr genes in these isolates, regardless of their pathogenic potential, represents public health and environmental concern. Agricultural practices, such as the use of litter, could facilitate the dissemination of mcr-carrying E. coli to other animals, humans, and the broader ecosystem. Similarly, a high proportion of E. coli isolates from market-ready birds carrying mcr genes may pose a direct risk to food safety for consumers.

The present study indicated that the broiler production chain was a reservoir for ESBL-producing E. coli that carries various ESBL genes, and there was a correlation between the AMR phenotypes and genotypes. This is in agreement with previous studies that reported high ESBL production of E. coli [66,68] in broiler production. This study found very high co-resistance between ESBL-producing E. coli and other classes of antimicrobials, including tetracycline, fluoroquinolones, aminoglycosides, trimethoprim-sulfamethoxazole, phenicols, and colistin, which is in line with the previous publications [69,70,71]. Overall, these findings highlight the complex relationships between different classes of antibiotics and the emergence and spread of ESBL-producing bacteria, underscoring the need for a comprehensive approach to antimicrobial stewardship and infection control.

The detection of bla_TEM and other resistance genes in most E. coli isolates suggests likely resistance to ampicillin and possibly amoxicillin–clavulanic acid, although confirmation requires sequencing, as not all variants confer ESBL activity. In contrast, bla_CTX-M is a true ESBL gene and signals resistance to third-generation cephalosporins. These findings underscore the importance of a One Health approach, emphasizing antimicrobial stewardship, reducing the use of HP-CIAs in animal production, and strengthening surveillance to limit the spread of resistance genes.

Implementing the recommended antimicrobial stewardship programs, reducing the use of HP-CIAs, and improving surveillance and biosecurity measures in the Indian poultry sector is feasible but may require a phased adoption approach due to resource limitations, varying farm sizes, and infrastructure variability. Targeted training programs, government support, and integration with existing production practices can facilitate the practical application of these measures, ultimately reducing the spread of antimicrobial-resistant E. coli along the broiler production chain.

This study is not without limitations. The diverse and complex nature of the broiler production chain in India, combined with limited resources, makes it challenging to achieve fully representative sampling. Additionally, variations in antimicrobial usage, farm management practices, biosecurity measures, good farming practices, and meat processing practices affect the inference of such studies.

5. Conclusions

This study demonstrated a high prevalence of antimicrobial resistance and resistance genes, including those conferring resistance to HP-CIAs, across the entire broiler production chain, from broiler breeder farms to the retail market and its environment. E. coli isolates exhibited extremely high to high resistance to most antimicrobials, with a strong correlation between AMR phenotypes and genotypes. Nearly all isolates were MDR, with 42% identified as ESBL producers, 65% resistant to third-generation cephalosporins, and 21% resistant to colistin. The presence of E. coli with resistance to HP-CIAs poses a significant public health risk, as these antimicrobials are critical for treating severe human infections. Furthermore, if these resistant E. coli isolates cause disease in chickens, treatment options would be severely limited, potentially leading to increased morbidity and mortality in poultry. Prudent antimicrobial use, effective biosecurity measures, and improved farm and meat processing management are crucial for reducing bacterial loads and minimizing the need for antimicrobial treatments. A comprehensive “One Health” approach, including strict antimicrobial stewardship, reduced reliance on HP-CIAs in animal production, and strengthened surveillance programs, is crucial to mitigating the spread of resistance genes across the broiler production chain.