Gram-negative gut colonization in children newly diagnosed with malignancy: microbiology, antimicrobial resistance pattern, and clinical implications

Abstract

Background

Carbapenem-resistant Enterobacterales (CRE) pose significant treatment difficulties owing to their high levels of antibiotic resistance. This study investigated the gut colonization rates of carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE) and extended-spectrum β-lactamase-producing (ESBL) gram-negative bacteria (GNB). Moreover, this study evaluated the antibiotic susceptibility profiles of CP-CRE and ESBL-producing Gram-negative bacteria. It also examines the potential infectious complications that may arise in pediatric patients recently diagnosed with cancer.

Methods

This prospective cohort study was conducted at a referral teaching oncology hospital in Shiraz, Iran. Fecal samples were collected at three time points: within 48–72 h of hospitalization, at the end of the first chemotherapy session, and during the next admission. The first sample identified cases of community-acquired colonization, whereas the second and third samples assessed hospital-acquired colonization by gram-negative bacteria. The infectious outcomes examined in this study included neutropenic enterocolitis (typhlitis), a life-threatening condition, and bloodstream infection.Antimicrobial susceptibility testing was performed for all gram-negative isolates using Clinical and Laboratory Standards Institute (CLSI)-standardized disk diffusion methods. Extended-spectrum β-lactamase (ESBL) production was screened and confirmed phenotypically, according to the CLSI criteria. In a subset of Escherichia coli (E. coli) isolates, polymerase chain reaction (PCR) was used to detect key β-lactamase and carbapenemase genes (blaCTX-M, blaTEM, blaSHV, blaOXA-48, blaNDM, and blaKPC).

Results

Eighty-one patients with a median age of 6.0 ± 0.49 years and a male: female ratio of 1.25 were included. 57 (70%) patients had community-acquired (CA) colonization with GNB, whereas eight (10%) had hospital-acquired (HA) colonization. E. coli and Klebsiella spp. were the most common pathogens in both colonization groups. In vitro susceptibility testing showed that colistin, amikacin, and carbapenems had the highest activity against the isolates. ESBL-producing GNB were detected in 91/157 (58%) fecal samples, and CP-CRE was detected in 23/157 (14.6%) samples. Twelve (21%) CA colonizers developed infectious outcomes.

Conclusions

The high incidence of CP-CRE colonization underscores the urgent need to monitor and control the spread of these highly resistant bacteria among newly diagnosed pediatric cancer patients in oncology settings.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13099-026-00812-w.

Background

The gut microbiota is a diverse group of microorganisms that live in the human digestive system and have a significant effect on human health and disease [1]. One crucial function of the gut microbiota is to prevent harmful microorganisms from colonizing the gastrointestinal tract. Changes in the gut microbiota, for example, due to antibiotic pressure, can alter its composition and weaken resistance against colonization, making the host more susceptible to pathogen colonization [2]. This is of particular concern for hospitalized patients, where antibiotic use and pathogen exposure are common.

Patients with hematologic malignancies are particularly susceptible to infections because of their immunocompromised status, neutropenia, and chemotherapy-induced mucosal barrier injury [3]. Chemotherapy-associated infections are among the main causes of morbidity and mortality in cancer, specifically multidrug-resistant (MDR) gram-negative (GN) bacteria [4], which have surpassed gram-positive infections in hospital-acquired settings [5]. Acinetobacter baumannii, Pseudomonas aeruginosa, Klebsiella pneumoniae, Escherichia coli (E. coli), and Enterobacter species are among the most common Gram-negative organisms associated with hospital-acquired infections in patients with cancer. MDR Gram-negative infections can result in treatment interruptions, prolonged hospital stays, higher healthcare costs, and lower survival rates [6], necessitating an understanding of their antibiotic sensitivity patterns and adequate empirical antimicrobial therapy to overcome these infections.

Globally, carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE) and extended-spectrum beta-lactamase-producing (ESBL) Enterobacterales have emerged as significant healthcare-associated pathogens because of their difficult management owing to their extended antibiotic resistance [7]. CP-CRE and ESBL can cause bloodstream infections (BSIs), urinary tract infections (UTIs), pneumonia, and surgical site infections (SSIs) [8]. Fecal colonization with ESBL and CP-CRE increases the odds of bacterial translocation, leading to further infections, particularly BSIs [9, 10]. Compared to ESBL Enterobacterales, CP-CRE colonization has been rarely investigated. Additionally, there are limited studies on CP-CRE in the pediatric population, especially in patients with cancer [11, 12].

Therefore, our study aimed to investigate the fecal colonization rate and antibiotic susceptibility of CP-CRE and ESBL GNB isolated from newly diagnosed pediatric cancer patients at a referral teaching hospital in Shiraz, Iran. We also investigated the complications associated with fecal colonization, including neutropenic enterocolitis (typhlitis) and bloodstream infections. In addition, we conducted molecular analyses to identify the genetic determinants of antibiotic resistance and phylogenetic background of colonizing E. coli strains.

Materials and methods

Study Design and Participants

We conducted a prospective cohort study among newly diagnosed pediatric cancer patients between October 15, 2022, and October 1, 2023, at Amir Oncology Hospital in Shiraz, Iran, after obtaining national ethical approval (ID: IR.SUMS.MED.REC.1401.333) [13].

The study included pediatric patients aged < 18 years who had received a cancer diagnosis and were admitted for their first course of immunosuppressive chemotherapy. Patients who met the inclusion criteria for fecal sampling and had at least one available stool specimen upon admission were analyzed. Patients who died of an underlying illness or experienced severe gastrointestinal (GI) adverse effects of chemotherapy, such as GI bleeding were excluded. The primary exposure of interest was the fecal gram-negative colonization status. Data on additional independent variables relevant to overall patient outcomes were collected for multivariate modeling. Intravenous antibiotic exposure, bloodstream infections, and neutropenic enterocolitis (typhlitis) were recorded at each admission.

Sample collection

The sampling procedure consisted of a three-step process to ensure a comprehensive evaluation of the patient’s fecal colonization status (Fig. 1). The first sample was obtained within the initial 48–72 h of the patient’s hospitalization (before starting any anticancer and antibiotic treatment), enabling the early investigation of community-acquired carriage. Antibiotic treatment was considered for critically ill patients suspected of having sepsis or bacteremia after a complete diagnostic workup (including pan culture) and stool sampling. A second sample was collected at the end of the first chemotherapy session, immediately before discharge. Finally, the third sampling was performed at the onset of the second hospitalization, within the first 48–72 h, or during subsequent hospitalizations in cases of bacteremia or neutropenic colitis. This approach facilitated a comprehensive assessment and provided a better understanding of the patient’s fecal colonization status. Stool samples were obtained from 81 patients at up to three sampling time points, yielding 157 stool samples. Each stool sample could yield one or more gram-negative bacterial isolates, and the same patient could contribute several isolates across repeated samples. Thus, the total number of bacterial isolates exceeded the number of stool samples.

Fig. 1.

Study flowchart of studied patients

Microbiology

All fecal or rectal swabs obtained from the patients were carefully transferred to the Clinical Microbiology Research Center in Cary Blair transport medium (Conda, Madrid, Spain) and kept at 4 °C for up to one day before testing. Upon arrival, the specimens were transferred to selective culture media to identify specific microbial strains of interest. The following media were used for culturing: (1) MacConkey agar plates (Biolife, Milan, Italy) supplemented with a meropenem disk (10 µg). Colonies within the carbapenem zone of inhibition were also examined. (2) Brain Heart Infusion (BHI) broth (Merck, Darmstadt, Germany) (5 ml) was enriched with a meropenem disk (10 µg), followed by overnight incubation at 35 °C, after which 100 µL of the broth culture was subcultured onto MacConkey agar for the isolation, enrichment, and separation of carbapenem-resistant strains. (3) MacConkey agar supplemented with 1 mg/L ceftriaxone (Sigma) was used to screen and isolate ESBL-producing species. Representative colonies from each selective media for further analysis, including carbapenem susceptibility testing, species-level identification, and identification of ESBL-producing strains in patient samples. Microorganisms were initially identified using biochemical tests. An API system (bioMérieux, France) was used for the final confirmation of the strains. To confirm ESBL production, Klebsiella spp. and E. coli isolates were subjected to a double-disc synergy test following the Clinical and Laboratory Standards Institute (CLSI) guidelines [14]. This test involved the use of ceftriaxone and ceftazidime alone and in combination with clavulanic acid. All CRE and ESBL strains were stored at − 80 °C in 40% glycerol broth medium for further analysis.

Antimicrobial Susceptibility Testing

Antimicrobial susceptibility testing was performed using CLSI-standardized methods, including disk diffusion for all antibiotics and broth microdilution for colistin susceptibility testing. Antibiotic susceptibility testing was performed on ESBL-producing and carbapenem-resistant Klebsiella spp. and E. coli samples using the disk diffusion method, as recommended by the CLSI-2022 guidelines. Commercially available antibiotic disks (Mast Co., Ltd., UK) were used according to the manufacturer’s instructions. The following antibiotics were used in the susceptibility assay: ampicillin (AMP 10 µg), ceftriaxone (CRO 30 µg), ceftazidime (CAZ 30 µg), cefepime (CPM 30 µg), ciprofloxacin (CIP 5 µg), levofloxacin (LEV 5 µg), amikacin (AK 30 µg), imipenem (IMI 10 µg), meropenem (MEM 10 µg), ertapenem (ETP 10 µg), gentamicin (GM 10 µg), cotrimoxazole (TS 25 µg), and piperacillin–tazobactam (PTZ 110 µg). Colistin susceptibility testing was performed using the broth microdilution method according to the CLSI criteria to determine the minimum inhibitory concentration (MIC). Antimicrobial susceptibility testing was quality-controlled using E. coli ATCC 25,922 in accordance with CLSI M100, and the results were accepted only when the QC values fell within the CLSI-approved ranges.

Molecular Detection of Resistance Genes and Phylogenetic Markers

ESBL- and carbapenem-resistance genes and phylogenetic markers were further assessed using polymerase chain reaction (PCR). Genomic DNA was extracted from confirmed E. coli isolates using standard boiling and column-based protocols. The presence of extended-spectrum β-lactamase (ESBL) and carbapenemase genes was assessed using multiplex and simplex PCR with specific primers targeting the following genes: blaCTX-M groups, blaTEM, blaSHV, blaOXA-48, blaNDM, blaKPC, blaVIM, and blaIMP. Primer sequences and cycling conditions were adopted from previously published protocols for ESBL and carbapenemase genes, as previously studied [15, 16]. Amplification products were visualized using agarose gel electrophoresis. For each PCR run, previously characterized clinical isolates harboring the respective target genes were included as positive controls, and nuclease-free water was used as a negative control, thereby validating the assay performance on each run.

To determine the phylogenetic group of each isolate, a triplex PCR assay was performed targeting chuA, yjaA, and TspE4.C2 genetic markers. Isolates were classified into groups A, B1, B2, or D based on the presence or absence of these markers, according to the Clermont method [17]. The gadA gene was amplified to confirm E. coli species identity.

Modified Carbapenem Inactivation Method (mCIM)

A 1-µL loopful of bacteria was resuspended in a 2-mL tube of TSB. A meropenem disk was added to each tube, and the tubes were incubated at 35 °C for 4 ± 15 min. The disks were then removed and placed on MH agar plates inoculated with a 0.5 McFarland suspension of carbapenem-susceptible E. coli ATCC 25,922. After incubating the plates at 35 °C for 16–20 h, the results were interpreted according to CLSI guidelines [14]. A zone size of ≥ 19 mm was considered negative, 6–15 mm was positive, and 16–18 mm with pinpoint colonies was also interpreted as positive (Fig. 2).

Fig. 2.

Modified Carbapenem Inactivation Method (mCIM) for Phenotypic Detection of Carbapenemase Production. mCIM results for the negative control and those positive for metallo-beta-lactamase-producing isolates

Definitions

All patients were diagnosed using criterion one (MBI-LCBI 1) from the Centers for Disease Control and Prevention’s National Healthcare Safety Network (NHSN) Patient Safety Component Manual, 2024 edition. Accordingly, mucosal barrier injury laboratory-confirmed bloodstream infection (MBI-LCBI) was diagnosed in patients whose blood cultures tested positive for gram-negative organisms included in the National Healthcare Safety Network MBI organism list (unrelated to an infection at another site) and who experienced neutropenia within a 7-day window encompassing the date when the positive blood culture was collected [18].

The classification and risk stratification of acute lymphoblastic leukemia (including T-cell leukemia) and lymphoma at our Children’s Oncology Center were guided by the protocols outlined in the 7th edition of Lanzkowsky’s Manual of Pediatric Hematology and Oncology, published in 2021 [19, 20].

Patients identified as carriers of gram-negative bacteria at the time of hospitalization were classified as community-acquired colonizers. In contrast, patients whose fecal samples tested positive for gram-negative bacteria 48–72 h after admission, provided they were not carriers upon admission, were classified as hospital-acquired colonizers. Incidental or short-term colonization refers to the transient presence of bacteria in a patient’s microbiota that does not lead to disease and may occur without any noticeable symptoms [21].

‘Colonization’ was defined as fecal carriage of antibiotic-resistant gram-negative bacteria (GNB), including ESBL-producing and/or carbapenemase-producing Enterobacterales, detected using selective antibiotic-containing media (MacConkey agar with ceftriaxone or meropenem enrichment). Growth on selective media, followed by confirmatory phenotypic tests and PCR for ESBL and carbapenemase genes, was considered evidence of resistant GNB colonization. ‘No growth’ was defined as the absence of ESBL- or carbapenemase-producing GNB on selective culture; this does not imply the absence of normal intestinal flora.

Multidrug resistance (MDR), extensively drug resistance (XDR), and pandrug resistance (PDR) were defined as acquired non-susceptibility to at least one agent in three or more antimicrobial categories (MDR), susceptibility to two or fewer categories (XDR), and non-susceptibility to all agents in all antimicrobial categories (PDR), according to the international ECDC/CDC consensus proposed by Magiorakos et al. and widely adopted in recent MDR Gram-negative literature [22–24].

Statistical analysis

The data were confirmed to be free of outliers and missing values. Continuous variables are reported as the mean ± standard deviation (SD) or standard error (SE) and median, depending on the normality of the data distribution. Categorical variables were compared using the χ² and Fisher’s exact tests for parametric and non-parametric distributions, respectively. Between-group comparisons of continuous variables were performed using the independent-samples t-test for parametric variables and the Mann-Whitney U test for non-parametric variables. Categorical variables were converted into dummy variables prior to the modeling. Univariate logistic regression was performed to calculate the odds ratios (ORs) and their corresponding 95% confidence intervals (CIs). Statistical significance was set at P < 0.05. An adjusted odds ratio was considered “clinically meaningful” when it fell within the range of < 0.50 or more than 2.00 [25–27]. Statistical analyses were performed using Stata^® software (version 17; StataCorp, Texas, USA).

Results

Sample characteristics

Of the 81 participants, 45 (56%) were men. The median age of the participants was 6.0 ± 0.49 y. The majority of patients were diagnosed with ALL (58 cases, 72%), followed by AML (10 cases, 12%). Less frequently diagnosed malignancies included non-Hodgkin’s lymphoma (4 cases, 5%), neuroblastoma (2 cases, 3%), glioblastoma (2 cases, 3%), rhabdomyosarcoma, sarcoma, brain stem glioma, Hodgkin’s lymphoma, and hepatoblastoma (1 case each, 1%).

Colonization status

Overall, 157 stool samples were collected from 81 patients (96 first-, 43 s-, and 18 third-samples). Culture yielded gram-negative bacteria in 119 samples, whereas 38 samples showed no growth. Of the 81 patients, 57 (70%) had community-acquired (CA) and eight (10%) had hospital-acquired (HA) colonization (Table 1). Among the 157 tested samples, E. coli and Klebsiella spp. were the most commonly identified pathogens in both CA and HA colonizers (Table 2).

Table 1.

Demographic and clinical characteristics and outcomes of 81 newly diagnosed pediatric cancer patients, comparing Non-colonizer with Community-acquired (CA) colonizers

Variables	Non-colonizer (24 cases)	CA colonizer (57 cases)	p. value
	N(%)	N(%)
Gender
Male	9 (37.5)	36 (63.16)	0.034†*
Female	15 (62.5)	21 (36.84)
Age Mean, (SE)	6.29 (0.87)	7.35 (0.59)	0.294††
Age (categorical)
<12 months	1 (4.2)	0 (0)	0.363†††
1–5 years	10 (41.7)	23 (40.4)
6–10 years	9 (37.5)	18 (28.1)
11–18 years	4 (16.7)	16 (31.6)
Malignancy type
Solid tumors	2 (8.33)	6 (10.53)	> 0.999†††
Hematologic malignancies	22 (91.67)	51 (89.47)
Baseline chemotherapy risk
Standard-risk	9 (37.5)	25 (43.8)	0.596†
High-risk	15 (62.5)	32 (56.1)
Disease relapse	2 (8.33)	7 (12.28)	0.718†††
History of antibiotic consumption a
No antibiotics	14 (58.33)	36 (63.16)	0.607†
1–2 antibiotics	5 (20.83)	14 (24.56)
>3 antibiotics	5 (20.83)	7 (12.28)
Infectious complications
FN	8 (33.33)	20 (35.09)	0.880†
FN episodes, Mean (SE)	0.625 (0.2)	0.578 (0.13)	0.956††
BSI	6 (25)	9 (15.79)	0.358†††
Neutropenic enterocolitis	2 (8.33)	7 (12.28)	0.718†††
Mucositis	5 (20.83)	8 (14.04)	0.447†
Prophylaxis
Ciprofloxacin	1 (4.17)	4 (7.02)	> 0.999†
SMX/TMP	18 (75)	42 (73.68)	> 0.999†
Amphotericin liposomal	8 (33.33)	20 (35.09)	0.880†
All-cause mortality	2 (8.3)	8 (14)	0.715†††

CA: Community-acquired, FN: febrile neutropenia; BSI: bloodstream infection; SMX/TMP:

a. History of antibiotic administration after the first sampling

* Statistically significant

† Chi-Square Test

†† Mann-Whitney Test

††† Fisher’s Exact Test

Table 2.

Microbiology of the colonized fecal Gram-negative pathogens (including ESBL*-producers and carbapenem-resistant isolates) during repeated sampling

Species	N**	1st sample N (%)	2nd sample N (%)	3rd sample N (%)
E. coli	104	65 (62.5)	29 (27.88)	10 (9.62)
Klebsiella spp.	11	3 (27.27)	3 (27.27)	5 (45.45)
Enterobacter spp.	3	3 (100)	-	-
Salmonella	1	1 (100)	-	-
No Growth	38	24 (63.16)	11 (28.95)	3 (7.89)
Total	157	96	43	18

* ESBL: extended-spectrum beta-lactamase-producing Enterobacterales

** Number of fecal samples

Susceptibility test results of isolated fecal Gram-negative bacteria

The overall sensitivity test results from the initial sampling indicated that among the recovered fecal isolates, ampicillin, ceftriaxone, cefotaxime, and SMX/TMP demonstrated the highest levels of resistance, with rates of 98.6%, 84.7%, 81.9%, and 81.9%, respectively. In contrast, colistin, amikacin, meropenem, ertapenem, and imipenem exhibited the highest sensitivity, with over 90% effectiveness against the isolates (Fig. 3). The antimicrobial sensitivity test results of repeated fecal sampling are shown in Figures S1-2. Pathogen-based antimicrobial sensitivity results indicated that the isolated E. coli strains had the highest sensitivity to colistin, amikacin, meropenem, ertapenem, and imipenem, with sensitivities of > 90% (Fig. 4). In contrast, these E. coli strains exhibited the highest resistance to ampicillin, ceftriaxone, cefotaxime, and SMX/TMP (98.5, 84.6, 83.1, and 83.1%, respectively). The overall antimicrobial sensitivity and resistance of the isolates closely mirrored those of E. coli isolates. This congruence underscores the significant impact of E. coli as the predominant fecal colonizer in the findings of this study. Figures S3-4 present the antimicrobial sensitivity test results for E. coli strains across repeated fecal samples.

Fig. 3.

Antimicrobial susceptibility test results of 96 Gram-negative isolates recovered from the first fecal samples. SMX/TMP: sulfamethoxazole/trimethoprim

Fig. 4.

Antimicrobial susceptibility test results for 65 E. coli strains recovered from the first fecal samples. SMX/TMP: sulfamethoxazole/trimethoprim

We also comprehensively evaluated and compared the resistance patterns among the widely used antibiotic classes. Most bacterial isolates exhibited a high resistance rate to third-generation cephalosporins (3rd GC), fluoroquinolones (FQ), and fourth-generation cephalosporins (4th GC) (89.1%, 72.3%, and 49.6%, respectively) (Figs. 5). Most E. coli isolates exhibited a high resistance rate to 3rd GC, FQ, and 4th GC (89.4%, 72.1%, and 46.2%, respectively). For Klebsiella spp. isolates, 3rd GCs, 4th GCs, FQs, AGs, and β-lactam/β-lactamase inhibitors exhibited the highest drug resistance rates (90.9%, 81.8%, 72.7%, 36.4%, and 27.3%, respectively). Compared to E. coli, Klebsiella spp. demonstrated higher resistance rates to 4th GC, AGs, and β-lactam/β-lactamase inhibitors (Figures S5).

Fig. 5.

Comparative analysis of antibiotic resistance rates between E. coli and K. pneumoniae spp. and all gram-negative isolates. Most bacterial isolates showed significant resistance to third-generation cephalosporins, fluoroquinolones, and fourth-generation cephalosporins. E. coli isolates also showed considerable resistance to these antibiotics, while Klebsiella spp. exhibited high resistance rates across multiple antibiotic classes, including third and fourth-generation cephalosporins, fluoroquinolones, aminoglycosides, and β-lactam/β-lactamase inhibitors. 3rd GC, Third-Generation Cephalosporins; 4th GC, Fourth-Generation Cephalosporins, FQ: Fluoroquinolones, MDR: Multidrug resistance; AG: Aminoglycosides, BL/BLI, β-lactam/β-lactamase inhibitors

Antibiotic resistance patterns of colonized pathogens

ESBL-producing GNB (ESBL-GNB) were detected in 52/96 (54.2%) specimens at first sampling. Of the second and third samples, 29/43 (67.4%) and 10/18 (55.6%) were ESBL-GNB, respectively. Third-generation cephalosporin-resistant non-ESBL-producing isolates (non-ESBL 3rd GCR) were recovered in 13/96 (13.5% ), 3/43 (7%), and 2/18 (11% ) of the 1st, 2nd, and 3rd of the collected samples, respectively. Overall, 58% (91/157) and 11.5% (18/157) of the samples were ESBL-producing GNB and non-ESBL 3rd GCR, respectively.

Carbapenemase-producing carbapenem-resistant Enterobacterales (CP-CRE) were detected in 16/96 (16.7%), 5/43 (11.6%), and 2/18 (11%) of the samples (1st, 2nd, and 3rd samples, respectively). In total, 23/157 (14.6%) of the recovered specimens were CP-CRE. Table 3 summarizes the results for ESBL-producing Gram-negative pathogens, CP-CRE, non-ESBL producers, third-generation cephalosporin-resistant species, and MDR gram-negative bacteria among CA and HA colonizers.

Table 3.

Antibiotic resistance profiles of fecal Gram-negative pathogens stratified by community-acquired (CA) and hospital-acquired (HA) colonization

variables	CA-colonizers N (%)	HA-colonizers N (%)
ESBL	52 (57.1)	39 (42.9)
CP-CRE	16 (69.6)	7 (30.4)
MDR	14 (50)	14 (50)

ESBL: Extended-spectrum β-lactamase-producing Gram-negatives; CP-CRE: carbapenemase-producing carbapenem-resistant Enterobacterales; MDR: multidrug-resistant Gram-negatives

Molecular Characterization of E. coli Isolates

To further characterize the antibiotic resistance profiles at the molecular level, 79 E. coli isolates were evaluated for the presence of resistance genes and phylogenetic markers. All isolates were confirmed to be E. coli by the presence of the gadA gene.

Among them, 76 isolates (96.2%) were identified as ESBL producers. The most frequently detected gene was blaCTX-M group 1 (blaCTX-M1), found in 63 isolates (79.75%), followed by blaTEM (50.6%) and blaSHV (10.1%) (Table 4). Several isolates carried multiple ESBL genes, with blaCTX-M1 and blaTEM co-occurring in 32 isolates, as shown in the gene co-occurrence heatmap (Fig. 6).

Table 4.

Frequency of resistance genes among E. coli isolates

Gene	Number (frequency)
Extended-spectrum beta-lactamase
blaCTX-M-1 group	63 (79.75)
blaTEM	40 (50.63)
blaSHV	8 (10.12)
blaCTX-M-8 group	4 (5.06)
blaCTX-M-2 group	3 (3.80)
blaCTX-M-9 group	3 (3.80)
blaCTX-M-25 group	2 (2.53)
Carbapenemase-producing
blaOXA-48	5 (6.33)
blaVIM	1 (1.26)
blaNDM	2 (2.53)
blaOXA-48 + NDM	2 (2.53)
blaKPC	0 (0)
blaIMP	0 (0)
Phylogenic marker
yjA	48 (60.76)
TSPE4.C2	39 (49.37)
ChuA	35 (44.30)

Fig. 6.

Frequency of ESBL- and Carbapenemase-Producing Resistance Genes

Carbapenemase genes were detected in 12.6% (10/79) of the E. coli isolates. Seven isolates carried blaOXA-48 gene. Four were positive for blaNDM, of which two co-harbored blaOXA-48. One isolate was blaVIM positive. None of the isolates tested positive for blaKPC or blaIMP .

The most common phylogroup was A (38%), followed by B2 (24%), D (20%), and B1 (18%). The predominance of phylogroup A suggests a substantial contribution from commensal strains, although the presence of B2 and D indicates the potential circulation of extra-intestinal pathogenic E. coli (ExPEC) clones, such as ST131 (Fig. 7).

Fig. 7.

Frequency of Phylogenetic Groups

Subsequent infectious complications

Among the 57 patients colonized with CA-GNB, 12 (21%) developed BSI or typhlitis during the observation period. In contrast, only one of the eight patients colonized with HA-GNB experienced subsequent infection complications.

The results of multivariable analysis

Factors associated with hospital-acquired GNB colonization

Patients aged 6–10-year-old age group, females, individuals with leukemia, and those experiencing cancer relapse had point estimates suggesting higher adjusted odds of HA-GNB colonization (aOR with 95% confidence interval [CI]: 2.98 [0.57–15.61], p = 0.95; 1.23 [0.18–8.08], p = 0.825; 2.41 [0.29–20.07], p = 0.415; and 2.09 [0.27–15.71], p = 0.473, respectively). Greater cumulative exposure to antibiotics was similarly associated with numerically higher odds of HA-GNB colonization (adjusted odds ratio [aOR] 1.34 [0.89–2.01], p = 0.158), whereas prophylactic liposomal amphotericin B and ciprofloxacin were associated with lower odds (aOR 0.07 [0.003–1.65], p = 0.101 and 0.82 [0.05–11.84], p = 0.886, respectively). However, all 95%CIs were wide and included 1.0, reflecting a modest number of events. Therefore, these estimates are best interpreted as exploratory and hypothesis-generating, with a clinical emphasis on the direction and magnitude of the associations rather than on formal statistical significance (Fig. 8).

Fig. 8.

Adjusted odds ratios (aORs) and 95% confidence intervals for factors associated with hospital-acquired Gram-negative bacterial colonization. Several confidence intervals are wide and cross 1.0, reflecting the limited number of events; estimates should therefore be interpreted as exploratory and hypothesis-generating

Factors associated with a bloodstream infection (BSI)

In multivariable analysis, several factors showed higher adjusted odds of BSI: age 11–18 years, female sex, solid tumor diagnosis, HA-GNB colonization, K. pneumoniae colonization, CP-CRE colonization, cancer relapse, and greater cumulative antibiotic exposure (aOR with 95%CI: 5.16 [0.90–29.42], p = 0.065; 4.64 [0.68–31.49], p = 0.116; 1.12 [0.10–12.28], p = 0.925; 1.29 [0.09–17.03], p = 0.843; 3.73 [1.25–11.04], p = 0.017; 2.81 [0.73–10.73], p = 0.130; 1.39 [0.14–13.63], p = 0.772; and 1.80 [1.06–3.04], p = 0.028, respectively). The strongest and most precise associations were observed for K. pneumoniae colonization and higher antibiotic exposure, whose CIs did not include 1.0, whereas most other predictors had wide CIs because of the limited number of BSI events. Figure 9 illustrates the direction and approximate magnitude of potential risk factors for BSI, which should be interpreted in conjunction with clinical plausibility and external evidence rather than solely on p-value thresholds.

Fig. 9.

Adjusted odds ratios (aORs) and 95% confidence intervals for factors associated with bloodstream infections. The forest plot illustrates the direction and relative magnitude of potential risk factors; wide confidence intervals, particularly for less frequent covariates, indicate imprecision and warrant cautious, clinically contextualized interpretation

Factors associated with a neutropenic enterocolitis (typhlitis)

In adjusted models, female sex, K. pneumoniae colonization, CP-CRE colonization, cancer relapse, greater exposure to antibiotics, and prophylaxis with ciprofloxacin or co-trimoxazole were associated with higher odds of typhlitis (aOR with 95%CI: 2.69 [0.50–14.27], p = 0.245; 5.97 [1.63–21.83], p = 0.007; 2.67 [0.49–14.44], p = 0.252; 2.15 [0.26–17.57], p = 0.474; 1.89 [1.09–3.29], p = 0.023; 1.50 [0.15–14.41], p = 0.723; 1.56 [0.31–7.90], p = 0.586; and 1.38 [0.19–9.98], p = 0.750, respectively). Patients aged 0–5 years and those with solid tumors had lower odds of typhlitis (aOR 0.91 [0.13–6.21], p = 0.925 and 0.39 [0.03–4.32], p = 0.445, respectively). While K. pneumoniae colonization and cumulative antibiotic exposure showed more precise estimates with CIs excluding 1.0, several other predictors had wide CIs that included the null hypothesis. These patterns suggest clinically plausible risk signals that should be interpreted cautiously and regarded as hypothesis-generating (Fig. 10).

Fig. 10.

Adjusted odds ratios (aORs) and 95% confidence intervals for factors associated with neutropenic enterocolitis (typhlitis). Because several confidence intervals are wide and some include 1.0, these estimates should be regarded as hypothesis-generating and interpreted alongside clinical plausibility and external evidence

As bloodstream and NE isolates were processed in the routine clinical microbiology laboratory and not retained for parallel research analyses, we did not perform strain-level matching between colonizing and infecting organisms.

Discussion

Gram-negative bacterial infections pose a significant global health threat, particularly in patients with malignancies such as leukemia. This study investigated the frequency and microbiology of GNB colonization in children newly diagnosed with cancer who were undergoing immunosuppressive chemotherapy. The findings revealed that 70% of the patients exhibited community-acquired (CA) colonization and 10% had hospital-acquired (HA) colonization, with HA colonizers showing higher rates of ESBL production and MDR than CA colonizers. In contrast, CP-CRE resistance was more prevalent among CA GNB colonizers. E. coli and Klebsiella spp. were the most prevalent fecal pathogens in our colonized patients, supporting their role in future infection-related events such as BSIs and typhlitis [28, 29].

The molecular resistance profiles and phylogenetic backgrounds of the colonizing E. coli isolates were characterized. Remarkably, 96.2% of the isolates were ESBL producers, with the blaCTX-M group 1 gene (most likely blaCTX-M-15) detected in approximately 80% of the samples, often co-occurring with blaTEM and blaSHV. This gene profile revealed a shift toward plasmid-mediated blaCTX-M dominance, highlighting the circulation of diverse and co-harbored resistance determinants in our setting.

Our expanded analysis demonstrated that, in addition to blaOXA-48, blaNDM was detected in four isolates, including two with blaOXA-48 co-carriage. The coexistence of blaNDM and blaOXA-48 mirrors regional epidemiological patterns and emphasizes the complexity of carbapenem resistance mechanisms in pediatric oncology patients The identification of NDM-producing isolates is clinically significant, given their association with limited therapeutic options and increasing prevalence across the Middle East. These molecular findings are consistent with the phenotypic resistance profiles and indicate that the intestinal microbiota of pediatric cancer patients harbors a substantial reservoir of multidrug-resistant E. coli, including lineages known to cause extra-intestinal infections in immunocompromised hosts.

The presence of carbapenem-resistant E. coli without identifiable carbapenemase genes suggests that porin loss or efflux mechanisms may also contribute to resistance [30, 31].

Our analysis showed a predominance of group A strains, which are typically considered commensals, followed by B2 and D groups, which include more virulent ExPEC clones, such as ST131. This distribution suggests that both low- and high-virulence E. coli lineages serve as reservoirs for ESBL and carbapenemase genes, potentially acquired through environmental or healthcare exposure. These findings have important clinical implications, as colonization with highly resistant E. coli (even from phylogenetically “benign” groups) may precede invasive infections in immunocompromised patients.

Because several models yielded wide confidence intervals owing to the limited number of outcome events, the associations between patient characteristics, HA-GNB colonization, bloodstream infection, and neutropenic enterocolitis should be interpreted cautiously. We therefore focus on the direction and approximate magnitude of the adjusted odds ratios, which are broadly compatible with biologically plausible mechanisms and previous reports linking intestinal MDR Enterobacterales colonization to subsequent gram-negative infections, rather than on strict p-value thresholds. Accordingly, these findings are exploratory and primarily intended to inform risk stratification and hypotheses for future larger multicenter studies rather than to provide definitive causal estimates.

Patients with cancer, particularly those with leukemia or cancer relapse, demonstrate an increased susceptibility to HA-GNB colonization, including E. coli [32]. Prior studies in oncology populations report a wide range of intestinal carriage of ESBL-producing Enterobacterales, from approximately 11% in high-risk hematologic malignancies in Germany to ~ 50–54% in adult [33] and pediatric cancer cohorts in Iran [34] and Algeria [35], with more recent data from Ethiopia showing even higher rates (> 80% in an oncology cohort) [36]. In community or healthy control groups, ESBL-E carriage is generally lower (often ~ 10–40%), although in some high-burden settings, it can be substantial; hospitalized or cancer patients consistently show higher carriage than healthy volunteers (for example 62.7% vs. 33.8% in India; 51% vs. 38% in Chad) [37–39]. More than half of the patients in this study were colonized with ESBL-producing GNB, a prevalence that was higher than that previously reported. Previous studies have indicated the persistence of ESBL-producing Enterobacterales carriage during long-term observation (> 35% of colonized individuals at 6 months), indicating that the carriage could be dynamically stable [40]. These colonized pathogens can result in bloodstream infections in patients with hematologic malignancies [41, 42], leading to approximately 80% of non-cancer-related deaths [43]. The mortality rates range from 18 to 30% and 21–30% for K. pneumoniae [44] and E. coli bacteremia, respectively, in this population [45–47], highlighting the negative impact of these pathogens.

Factors contributing to the higher probability of colonization in patients with cancer include low intestinal microbiota diversity [48, 49], neutropenia, prior broad-spectrum antibiotic treatment, and gastrointestinal mucosal injuries [49–51]. Previous fecal carriage of gram-negative bacteria also increases the likelihood of recurrent colonization [51]. According to our findings, leukemia, cancer relapse, increased antibiotic exposure, and frequent febrile neutropenic episodes were associated with a trend toward higher odds of GNB colonization by HA. Moreover, liposomal amphotericin B and ciprofloxacin for preventive purposes appeared to reduce the risk of HA-GNB colonization (Fig. 9). Although these trends were clinically plausible, they did not reach statistical significance and should be interpreted cautiously.

In our study, HA-GNB colonization was significantly associated with a higher risk of BSI. This stronger association for HA colonization likely reflects that children who acquire MDR-GNB during hospitalization represent a higher-risk subgroup with longer admissions, more intensive chemotherapy, prolonged or profound neutropenia, invasive devices, and recent exposure to broad-spectrum antibiotics—all recognized risk factors for both the acquisition of carbapenem-resistant or ESBL-producing Enterobacterales and subsequent bacteremia in hematology–oncology populations. In addition, hospital-acquired colonization often involves highly resistant, hospital-adapted lineages, making translocation across damaged mucosa more likely to result in concordant MDR-GNB bloodstream infections.

Additionally, increased antibiotic exposure was associated with significantly higher odds of developing BSI and neutropenic enterocolitis. Similarly, a study on allogeneic hematopoietic cell transplant recipients found that the probability of GNB BSI (specifically MDR-GNB and ESBL-producing Enterobacterales) was significantly higher in patients colonized with these pathogens who received fluoroquinolone prophylaxis during neutropenia [52]. In another study, β-lactam/β-lactamase inhibitor combinations and exposure to carbapenems were identified as risk factors for HA K. pneumoniae BSI [53]. Additionally, younger age, exposure to carbapenems or aminoglycosides, neutropenia, transmission of serine carbapenemase-producing bacteria (especially KPC), colonization by multiple carbapenemase-producing Enterobacterales, and detection of multiple carbapenemase genes have been reported to be associated with CPE-BSI [54]. These findings highlight the role of gut colonization by pathogens in BSI and suggest that antibiotic treatment may not benefit all colonized patients. In our cohort, Klebsiella spp. exhibited broader resistance patterns than E. coli, particularly to fourth-generation cephalosporins, aminoglycosides, and β-lactam/β-lactamase inhibitor combinations (Fig. 11), underscoring the therapeutic challenges posed by these pathogens in pediatric oncology.

Fig. 11.

Comparative assessment for antibiotic resistance rate between E. coli and K. pneumoniae spp. obtained from fecal samples. Third_GC_RESIST: Third-Generation Cephalosporin resistant; Fourth_GC_RESIST: Fourth-Generation Cephalosporin resistant, FQ_RESIST: Fluoroquinolone resistant; MDR: Multidrug resistant; AG_RESIST: Aminoglycoside resistant; BL/BLI_RESIST: β-lactam/β-lactamase inhibitor resistant; COLISTIN_RESIST: Colistin resistant

Recent advances in β-lactam/β-lactamase inhibitor combinations and siderophore cephalosporins have provided new therapeutic options for highly resistant Enterobacterales. Aztreonam–avibactam and cefiderocol have shown potent activity against blaNDM-producing E. coli isolates [55], overcoming the limitations of carbapenems and older β-lactams. Similarly, ceftazidime/avibactam and plazomicin have demonstrated efficacy against carbapenem-resistant K. pneumoniae and E. coli, offering combination strategies that may suppress the emergence of resistance [56]. Fosfomycin has also re-emerged as a valuable adjunct, showing activity against ESBL- and blaNDM-5-producing E. coli and K. pneumoniae clinical isolates [57]. The integration of these newer agents into empiric or targeted regimens, guided by molecular diagnostics, could help mitigate mortality from MDR infections in oncology settings, particularly in regions with endemic NDM and OXA-48 producers.

Controversies exist in the literature regarding the control of MDR GN bacterial transmission. Current evidence suggests that bundled interventions are more effective than single interventions. Although there is a lack of standardized approaches to control MDR GN bacteria, contact precautions, single-room isolation, active surveillance cultures, environmental cleaning, disinfection, staff and patient cohorts, healthcare worker screening, and patient decolonization are accepted evidence-based practices and preventive measures [58].

Our findings have practical implications for infection prevention and empirical treatment in pediatric oncology. The high baseline rate of gut colonization with ESBL- and carbapenem-resistant Enterobacterales suggests that targeted rectal screening may be considered for children expected to have prolonged or profound neutropenia, particularly in high-prevalence settings, to support infection control and risk stratification [59–61]. However, current pediatric fever and neutropenia guidelines do not recommend routine escalation of initial empirical therapy based on colonization status alone; instead, colonization data should be interpreted together with clinical severity, with early broadening reserved for hemodynamically unstable patients or when the colonizing MDR organism is strongly suspected as the cause of sepsis [62, 63]. In line with guideline recommendations, fluoroquinolone prophylaxis should be restricted to carefully selected high-risk subgroups and is unlikely to be appropriate where baseline fluoroquinolone resistance in gut Enterobacterales is high, as in our cohort [64].

This study had several methodological limitations that should be considered when interpreting our findings. The limited sample size of pediatric patients aged < 18 years may affect the generalizability of our findings. The number of BSI and neutropenic enterocolitis episodes was relatively small compared with the number of covariates included in our multivariable models, which may have contributed to overfitting and wide confidence intervals; consequently, the estimated associations should be interpreted as exploratory and hypothesis-generating.

Finally, we did not collect or sequence BSI/typhlitis isolates in parallel with the colonizing stool isolates and therefore could not determine genetic or detailed phenotypic concordance between colonizing and infecting strains. As a result, our data support an association between MDR gram-negative gut colonization and subsequent BSI/typhlitis rather than proving that specific colonizing strains caused these infections.

As a prospective baseline study of gram-negative intestinal colonization in children at the time of cancer diagnosis, our work integrates detailed phenotypic susceptibility patterns with the molecular characterization of ESBL and carbapenemase genes in a middle-income, high-antimicrobial-resistance setting. By linking colonization profiles with early infectious outcomes, these data complement and extend previous evidence that multidrug-resistant gram-negative gut colonization is a key risk factor for subsequent invasive infection in hematology–oncology patients. These findings provide locally grounded information that can support risk-stratified empirical antibiotic choices and inform antimicrobial stewardship and infection prevention strategies in pediatric oncology units with similar epidemiology.

Conclusion

Our findings demonstrate a high burden of intestinal colonization with extended-spectrum β-lactamase-producing and carbapenemase-producing Enterobacterales—predominantly Escherichia coli and Klebsiella spp.—among newly diagnosed pediatric cancer patients and suggest that such colonization is associated with subsequent bloodstream infection and neutropenic enterocolitis. In line with data showing that ESBL/CP-Enterobacterales carriage increases the risk of invasive infection in oncology populations, our results support the implementation of context-adapted active surveillance, strengthened infection prevention bundles, and rigorous antimicrobial stewardship in pediatric oncology units, especially in high-burden settings. In conclusion, the high baseline prevalence of multidrug-resistant gram-negative gut colonization at cancer diagnosis in our cohort supports targeted admission screening of children expected to experience prolonged neutropenia and stewardship strategies that use colonization status to both prompt timely escalation in unstable colonized patients and safely avoid unnecessary carbapenem use in stable non-colonized children.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Study concept and design: AA, AD, FS, and BP; Acquisition of data: AA, AD, FS, and BP; ‎Statistical Analysis: AA, BP, HMV, and SRAM, Analysis and interpretation of data: AA, AD, FS, and BP; Drafting of the ‎manuscript: AA, SS, SA, SS, AM, and SRAM, Critical ‎revision of the manuscript for important intellectual ‎content: AA, BP, SRAM, EM, and ‎HMV; Study supervision: AA, PB, and HMV. All individuals listed as ‎‎(co)-authors ‎have met the authorship criteria, and no one who qualifies for ‎authorship is ‎omitted from the list. The final manuscript was corrected ‎and approved by all ‎authors.‎.

Funding

Not applicable.

Data availability

The dataset used and analyzed during the current study will be available ‎‎from the corresponding author upon reasonable request.‎.

Declarations

Ethics approval and consent to participate

The study was designed to comply with the national norms and regulations for conducting medical research in Iran, as well as ethical principles. The study was approved by the “Iran National Committee for Ethics in Biomedical Research” with approval ID: IR.SUMS.MED.REC.1401.333 [13]. All of the participants or their parents were informed about this study, and the parents who allowed their children to be tested signed a written informed consent form.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.