Risk factors and molecular epidemiology of intestinal colonization by carbapenem-resistant Gram-negative bacteria in patients with hematological diseases: a multicenter case‒control study

ABSTRACT Patients with hematological diseases are considered to be at high risk for intestinal colonization by carbapenem-resistant Gram-negative bacteria (CR-GNB). However, the epidemiological data regarding risk factors and molecular characteristics of intestinal colonized CR-GNB isolates in this population are insufficient in China. A multicenter case‒control study involving 4,641 adult patients with hematological diseases from 92 hospitals across China was conducted. Following culture of collected rectal swabs, mass spectrometry and antimicrobial susceptibility tests were performed to identify GNB species and CR phenotype. Risk factors were assessed through retrospective clinical information. Whole-genome sequencing was used to analyze the molecular characteristics of CR-GNB isolates. This trial is registered with ClinicalTrials.gov as NCT05002582. Our results demonstrated that among 4,641 adult patients, 10.8% had intestinal colonization by CR-GNB. Of these, 8.1% were colonized by carbapenem-resistant Enterobacterales (CRE), 2.6% were colonized by carbapenem-resistant Pseudomonas aeruginosa (CRPA), and 0.3% were colonized by carbapenem-resistant Acinetobacter baumannii (CRAB). The risk factors for CR-GNB colonization include male gender, acute leukemia, hematopoietic stem cell transplantation, β-lactam antibiotic usage, and the presence of non-perianal infections within 1 week. Compared with CRPA-colonized patients, patients using carbapenems were more likely to be colonized with CRE. NDM was the predominant carbapenemase in colonized CRE. This study revealed a high CR-GNB intestinal colonization rate among adult patients with hematological diseases in China, with CRE being the predominant one. Notably, a significant proportion of CRE exhibited metallo-β-lactamase production, indicating a concerning trend. These findings emphasize the importance of active screening for CR-GNB colonization in patients with hematological diseases. IMPORTANCE Carbapenem-resistant Gram-negative bacteria (CR-GNB) has emerged as a significant threat to public health. Patients with hematological diseases are at high risk of CR-GNB infections due to their immunosuppressed state. CR-GNB colonization is an independent risk factor for subsequent infection. Understanding the risk factors and molecular characteristics of CR-GNB associated with intestinal colonization in patients with hematological diseases is crucial for empirical treatment, particularly in patients with febrile neutropenia. However, the epidemiology data are still insufficient, and our study aims to determine the intestinal colonization rate of CR-GNB, identify colonization risk factors, and analyze the molecular characteristics of colonized CR-GNB isolates.

Patients with hematological diseases are at high risk of CR-GNB infection due to factors such as bone marrow suppression, gastrointestinal mucositis resulting from the use of immunosuppressants and chemotherapy drugs, prolonged hospitalization, neutrophil deficiency, and frequent use of broad-spectrum antibiotics (7).Previous research has identified intestinal colonization by CR-GNB as an independent risk factor for subsequent infection.In cases where the colonized CR-GNB strains breach the intestinal mucosal barrier, severe infections can occur, with mortality rates ranging from 65% to 100% (8,9).Furthermore, studies have shown that intestinal colonization by drug-resistant bacteria can impact the overall survival and increase mortality rates in patients undergoing hematopoietic stem cell transplantation (HSCT) (10).Consequently, active screening for intestinal CR-GNB colonization in patients with hematological diseases has become a crucial preventive measure for controlling and averting nosoco mial infections (11,12).
The rates of CR-GNB intestinal colonization in patients with hematological diseases vary across different regions worldwide.For instance, an Italian multicenter observational study reported a CR-GNB intestinal colonization rate of 3.8% in patients with hematolog ical malignancies (13).However, currently, there is a lack of epidemiological data on CR-GNB intestinal colonization in patients with hematological diseases in China.This study aims to address this gap by conducting a multicenter observational study; in this study, we determined the rate of CR-GNB intestinal colonization, identified risk factors for colonization, and analyzed the molecular characteristics of colonized CR-GNB isolates in China.The findings from this study contribute to reducing the colonization rate and the incidence of infection and mortality associated with CR-GNB in patients with hematological diseases.

Intestinal colonization rate of CR-GNB in patients with hematological diseases
From July 20, 2021, to December 31, a total of 5,053 non-duplicated inpatients from 92 hospitals in China were screened, and 4,641 adult patients (age ≥18) were analyzed as shown in Fig. 1.The number of patients screened from each province is shown in Fig. 2A.Out of 4,641 adult patients, 499 were found to have intestinal colonization with CR-GNB (including CRE, CRPA, or CRAB) isolates, and the colonization rate was 10.8% (499/4,641).Further analysis of the intestinal colonization patients revealed that 376 patients had CRE colonization, 121 had CRPA colonization, and 15 had CRAB colonization.In 448 patients, only one strain was isolated, while in the other 51 patients, two strains were isolated, and the CRE, CRPA, and CRAB intestinal colonization rates in adult patients with hematological diseases were 8.1% (376/4,641), 2.6% (121/4,641), and 0.3% (15/4,641), respectively.
Additionally, the intestinal CR-GNB colonization rate was different in different provinces, as shown in Fig. 2B.The colonization rate was highest in Beijing and Shanghai and lowest in Qinghai and Tibet.Additionally, we analyzed the CRE and CRPA coloniza tion rates in seven geographical regions in China.As shown in Fig. 2C, the CRE coloniza tion rate was higher in East, South, and North China, while the CRPA colonization rate was higher in North and Central China.In total, the intestinal CR-GNB colonization rates in the southeastern and northeastern parts of China were significantly higher than those in the southwestern and northwestern parts.

Risk factors for CR-GNB intestinal colonization in adult patients
To identify the risk factors for intestinal colonization by CR-GNB isolates, patients were categorized into either the colonization group or the non-colonization group based on the presence or absence of intestinal colonization.Overall, 499 colonized and 4,142 non-colonized patients were included in the analysis.The demographics and character istics of patients with hematological diseases before screening are shown in Table 1.Both groups showed a male predominance.Acute leukemia accounted for the highest proportion of all hematological diseases.
Additionally, we compared the risk factor differences between CRE-and CRPA-colon ized patients.As shown in Table S1, 376 CRE-colonized patients and 121 CRPA-colonized patients were included in analysis, and the multivariate analysis showed that compared with CRPA-colonized patients, patients using other β-lactams (P < 0.001) or carbapenems and other β-lactams (P = 0.006) were more likely to be colonized with CRE isolates.(Table S2)
For CRPA isolates, all the antipseudomonal agents exhibited favorable antimicrobial activity, with resistance rates less than 30%.Moreover, in CRAB isolates, only polymyxin B and tigecycline showed low resistance rates.

DISCUSSION
This multicenter case-control study was conducted in China and involved specimens from 31 provinces.The results revealed that the intestinal colonization rate of CR-GNB in adult patients with hematological diseases in China was 10.8%.Comparing these findings with reports from European countries, the colonization rate in China was higher than the reported rate of 3%-4% in Europe (12,13).However, the rate in China was lower than the rate of 75.5% reported in India (14).Additionally, single-center studies in China have reported CRE colonization rates ranging from 6.6% to 23.8% in patients with hematologic malignancies (15,16).Among all colonized CR-GNB isolates, CRE coloniza tion accounted for the highest, followed by CRPA, while CRAB colonization was rare.This was different from ICU patients, in whom CRKP and CRAB were the most commonly colonized pathogens (17,18).Moreover, metalloenzymes, especially NDM, produced in CRE isolated from patients with hematological diseases were significantly higher than the isolates separated from the whole population.In our study, the portion of NDM was 79.9% and 76.2% in CREC and carbapenem-resistant Enterobacter spp., respectively, much higher than the result of resistance surveillance from 2015 to 2016 in China, which reported that NDM was the predominant carbapenemase in CREC (52.1%) and carbapenem-resistant Enterobacter spp.(50.3%) (19).In CRKP, although KPC was still the predominant carbapenemase (35.1%), the portion of NDM (14%) elevated signifi cantly compared to the isolates from the ICU (20,21).CRKP strains also had a high percentage of non-carbapenemase-producing strains (up to 48%).The carbapenem MICs of the non-carbapenemase-producing strains were lower than those of carbapenemaseproducing strains.These strains were all producing ESBL enzymes or AmpC enzymes, and the possible carbapenem resistance mechanism for these non-carbapenemase-pro ducing strains may be high production of ESBLs and AmpC enzymes together with inactivation of pore proteins.This study also found that the colonization rate fluctu ated in different regions in China, ranging from 0% to 20.0%.These findings highlight the need for CR-GNB intestinal colonization screening in patients with hematological diseases in each region.Epidemiology data can aid in selecting appropriate empirical treatments and implementing measures to control the spread of CR-GNB isolates.
Identifying the risk factors for CR-GNB colonization can aid in determining whether patients are at high risk, which enables healthcare facilities to selectively implement proactive screening measures.However, studies on the risk factors for CR-GNB coloniza tion in patients with hematological diseases are inadequate in China.Cao et al. found that pulmonary infection, perianal infection, and carbapenem application in the 3-month pretransplant were independent risk factors for CRE colonization in HSCT recipients (16).Jaiswal et al. found that acute leukemia and previous long-term hospitalization were independent risk factors for CRE colonization in patients with hematological malignancies in a cohort study (22).In an Italian multicenter study, less age (< 60 years), non-autologous HSCT, and urinary tract catheterization were risk factors for CR-GNB intestinal colonization in patients with hematological diseases (13).Our study showed that in adult patients with hematological diseases, male sex (P = 0.045), acute leukemia (P = 0.048), HSCT (P = 0.011), carbapenem (P = 0.017), and/or other β-lactam (P = 0.009) usage in the 6 months before screening and other infections within 1 week before screening (except for perianal infection) (P = 0.008) were risk factors for CR-GNB intestinal colonization.Most patients with acute leukemia need HSCT therapy, and HSCT can cause mucosal toxicity and leads to increased permeability in the gastrointestinal tract, prolonged hospitalization, and use of antibiotics (7).Previous infection also leads to antibiotic usage.Antibiotic usage may disrupt the gastrointestinal microflora and eradicate susceptible competing strains, thus promoting CR-GNB colonization.Moreover, our study compared the risk factor difference between CRE-and CRPA-colonized patients.The result showed using of other β-lactams or carbapenem and other β-lactams was more likely to cause CRE intestinal colonization.A previous study reported that reintroduction of E. coli in antibiotic-treated mice increases defense against P. aeruginosa colonization (23), while the interaction between CRE and CRPA in the intestine needs to be further explored.
Furthermore, our study also clarifies the molecular characteristics of CR-GNB isolates.Phylogenetic analysis results showed the extreme divergence of CREC and CRKP strains from patients with hematological diseases.Clonal transmission could be observed, but unlike CRE strains isolated from the general population, the proportion of traditional dominant clones, such as ST11, among CRKP isolates decreased significantly (19,24).This may suggest that original Enterobacterales strains colonized in the intestine undergo evolution under the selective pressure of antimicrobial agents.And, in recent years, the rate of carbapenemase-producing CRPA has gradually increased (20).Globally, the most common carbapenemases in CRPA are metalloenzymes, including VIM, IMP, and NDM (21).However, some studies in Zhejiang Province reported that KPC-2-producing CRPA accounted for 21.1%-40.4% of all isolates, while ST463-CRPA accounted for 43.8%-70.9% of KPC-2-producing isolates (25,26).In our study, the proportion of carbapene mase-producing CRPA was 5.0% (6/121).Notably, five of the isolates produced KPC-2 and belonged to ST463; these strains were all isolated from Zhejiang Province.Recent studies showed that KPC-2-producing ST463-CRPA carries both type III secretion systems, ExoU and ExoS, indicating the coexistence of resistance and hypervirulence and may be found in emerging high-risk clones (27,28).
There are still some limitations to this study.First, although this study included specimens that were collected across the country, the number of specimens differed in each province, which may have resulted in bias.Second, although this study included many patients, certain variables, such as neutropenia and previous ICU admission, were not evaluated.These factors may also influence the assessment of risk.
In conclusion, in this study, we analyzed the risk factors for intestinal colonization by CR-GNB isolates in patients with hematological diseases by performing a multicen ter case-control study.The CR-GNB intestinal colonization rate was high in patients with hematological diseases in China, among which CRE was the main pathogen.The proportion of colonized CRE isolates with metallo-β-lactamases was high, followed by CRPA, while the CRAB colonization was rare.Thus, active screening is important for patients with hematological diseases.

Study design and participants
This was an observational multicenter study conducted from July 20, 2021, to Decem ber 31, 2021, in 92 hospitals across China.This study aimed to screen inpatients with hematological diseases for intestinal colonization by CR-GNB.Rectal swabs or feces were collected from all inpatients of the hematology ward at certain hospitals on specific days (Table S10), with no age or diagnosis exclusion.As shown in Fig. 1, a total of 4,641 adult patients were included.
All samples were sent to the central laboratory and cultured in trypticase soy broth (TSB) at 37°C overnight.The following day, the cultured TSB samples were inoculated on specific media for the identification of different species, including E. coli/ coliform, Acinetobacter chromogenic media (CHROMagar, France), P. aeruginosa selection medium (Sangon Biotech, China), and K. pneumoniae selection medium (Hopebio, China).Monoclonal colonies that were selected were further analyzed using MicroTyper MS (Skyray Instruments, China) to determine the exact species.Confirmed monoclonal colonies were saved for carbapenem susceptibility testing.

Statistical analysis
The clinical data of all patients were collected retrospectively, including (1) general information such as age and gender and (2) risk factors including the diagnosis of hematological disease, time of illness, treatment within 6 months (chemotherapy, HSCT, or only symptomatic and supportive treatment), antimicrobial agent use within 6 months, diabetes, deep vein catheterization, gastrointestinal illness symptoms (abdomi nal pain, diarrhea, and gastrointestinal bleeding) within 1 week, and signs of infections (blood, perianal, abdominal cavity, or other parts) within 1 week.
Data were analyzed using SPSS 26.0 (IBM Corporation).Binary logistic regression analysis was used, and the results included odds ratios (ORs) and 95% confidence intervals (CIs).Variables with P < 0.2 in univariate analysis were included in multivariate analysis, and P < 0.05 was considered statistically significant.The GPower (v3.1.9.7) was used to do a post hoc power analysis.And t-test and linear multiple regression fixed model was used.The total sample size was 4641, the α error was 0.05, and the effect size was 0.15.

FIG 1
FIG 1Flowchart showing the recruitment procedure, risk factor analysis of CR-GNB intestinal colonization, and molecular characteristic analysis of CR-GNB isolates.

FIG 2 (
FIG 2 (A) The number of screened individuals in various provinces and autonomous regions of China.(B) The intestinal CR-GNB colonization rate in each province.A darker color represents a greater number of screened patients or a higher colonization rate.(C) The intestinal CRE, CRPA, and CR-GNB colonization rates in each administrative region of China.Red represents the CRE colonization rate; green represents the CRPA colonization rate; and the line segment represents the CR-GNB colonization rate.The maps were generated from the Standard map service website provided by the Center for Mapping Technology Review, Ministry of Natural Resources, People's Republic of China.

FIG 3
FIG 3 Carbapenemase distribution in CRE isolates.Gray represents non-carbapenemase isolates.Yellow represents KPC-producing isolates.Brown represents isolates producing both KPC and NDM.Light blue represents NDM-producing isolates.Pink represents IMP-producing isolates.Dark blue represents isolates producing both NDM and OXA.Purple represents isolates producing both NDM and IMP. Green represents OXA-producing isolates.The number represents the exact portion of each carbapenemase in each CRE species.

FIG 4
FIG 4 Phylogenetic population structures: (A) 184 CREC isolates.(B) 171 CRKP isolates.The phylogenetic trees were rooted in the midpoint.MLST, multilocus sequence type.In circle 1, the unlabeled isolates represent scattered STs (n < 5) (see supplementary tables).Circle 2 represents the region from which patient isolates were collected.Circle 3 represents the carbapenemase type produced by the isolates.

TABLE 1
Characteristics of patients with hematological diseases

TABLE 2
Risk factors for intestinal colonization of CR-GNB in patients with hematological diseases

TABLE 3
Antimicrobial susceptibility results of CR-GNB isolates .