Clinical Analysis of Bloodstream Infection of Escherichia coli in Patients with Pancreatic Cancer from 2011 to 2019

Background Pancreatic cancer patients were particularly predisposed to develop Escherichia coli (E. coli) bloodstream infection (BSI); however, little information is currently available. We set out to find E. coli BSI's risk factors in pancreatic cancer to provide valuable experience. Methods We retrospectively analyzed the clinical data of pancreatic cancer patients (31 cases with E. coli BSI and 93 cases without BSI) by a case-control study. SPSS 17.0 was adopted to perform univariate and multivariate analyses. Bacterial resistance analysis was performed by Whonet 5.6. Results Hospitalization days ≥7 days, number of admissions ≥2 times, surgery, chemotherapy, the type of antibiotics used ≥2 species, albumin<40.0 g/L, and prealbumin < 0.2 g/L were the potential risk factors for pancreatic cancer patients with E. coli BSI (P < 0.1). Multivariate logistic regression showed hospitalization days ≥7 days (OR = 11.196, 95% CI = 0.024–0.333, P < 0.001), surgery (OR = 32.053, 95% CI = 0.007–0.137, P < 0.001), and chemotherapy (OR = 6.174, 95% CI = 0.038–0.688, P=0.014) were the independent risk factors for E. coli BSI of pancreatic cancer patients. E. coli resistant to carbapenems was rare; they were susceptible to cephamycin and piperacillin/tazobactam. The 90-day mortality rate of the infected group was significantly higher than the control group (41.9% versus 8.6%, P < 0.001). Conclusions Hospitalization days ≥7 days, surgery, and chemotherapy are the independent risk factors for E. coli BSI of pancreatic cancer patients, which allows us to identify patients at potential risk and perform preventive treatment in time.


Introduction
Bloodstream infection (BSI) has a high mortality rate and is costly to treat; it is the most challenging type of infection to control and has a significant impact on the prognosis of patients [1]. Due to the improper use of antibiotics, pathogenic bacteria's resistance to commonly used antibacterial drugs is increasing [2], coupled with the adverse reactions caused by the abuse of antibiotics [3], which has made the treatment of BSI difficult.
In patients with malignant tumors, due to integrity damage (e.g., surgery, trauma, and indwelling catheters) and some medical sources (e.g., antibiotic use, radiotherapy, and chemotherapy) or nonmedical factors (such as advanced age and poor nutritional status), the body's immunity is reduced or dysfunctions, and thus can easily induce various infections [4]. ere are differences in the distribution and drug resistance of pathogens in BSI in different cancer patients [5]. We analyzed the BSI data of Tianjin Medical University Cancer Institute and Hospital (TMUCIH) for nine years and found that the most common pathogen of BSI in cancer patients was Escherichia coli (E. coli), and pancreatic cancer patients rank as the first of E. coli BSI.
E. coli is the leading cause of BSI involving Gramnegative bacteria [6]. e last 20 years have witnessed a striking increase of BSI caused by antibiotic-resistant isolates of E. coli [7,8]. Studies have shown that failure to provide timely and effective antibacterial therapy for BSI caused by extended-spectrum-lacta-mases (ESBL)-producing E. coli is associated with increased mortality [7,9].
Pancreatic cancer is a malignant tumor of the digestive tract with high malignancy and difficulty in diagnosis and treatment [10,11]. Postoperative adjuvant chemotherapy such as radiotherapy and chemotherapy can improve the survival rate; however, the 5-year survival rate is only around 5%, which is still one of the worst prognostic malignancies [10,12]. Early diagnosis and treatment are key to improve the prognosis of pancreatic cancer [13].
ere are 2,600 beds in TMUCIH, more than 80 beds in the hepatobiliary and gastrointestinal oncology departments, and only 30 beds in the pancreatic oncology department. However, the proportion of E. coli BSI was the highest, of which the cause remained to be determined. Further analysis of the risk factors can assist in the diagnosis and treatment of pancreatic cancer patients with BSI. For various reasons, patients with solid tumors are particularly predisposed to develop BSI; however, little information is currently available, let alone pancreatic cancer. erefore, our data are precious and can provide valuable experience.

Patients and Study Design.
Our study analyzed 2386 blood culture-positive pathogens of cancer patients from January 2011 to December 2019 in TMUCIH (exclude duplicate isolates isolated from the same patient). e first ranked pathogen was E. coli (697 cases). e primary cancer patients with E. coli BSI were pancreatic cancer patients (100 cases). 31 pancreatic cancer patients with E. coli BSI were randomly selected and defined as the infected group, and 93 pancreatic cancer patients without BSI (no blood culture was drawn during hospitalization and have no episodes of Gramnegative bacteremia), sex, age, and admission time matched with infected group, were collected as the control group. All cancer types in this study were confirmed by pathological diagnosis. e diagnosis of pancreatic cancer was made prior to hospitalization with E. coli BSI or made during the same hospitalization but prior to the onset of E. coli BSI.
e clinical data of the infected and control groups were retrospectively analyzed by a case-control study. e clinical data, including gender, age, the number of hospitalization days, clinical stage, the number of admissions, surgery, intraoperative bleeding, chemotherapy, radiotherapy, the type of antibiotics used, distant organ metastasis of cancer, invasive operation (central venous catheter, drainage tube, or urine tube), merger with other parts of infection, stayed in ICU, blood transfusion, combined with diabetes, glutamine transfusion, white blood cells <4.0 × 10 9 /L, neutropenia (neutrophil<1.5 × 10 9 /L), albumin<40.0 g/L, and prealbumin < 0.2 g/L, were collected. All of the above factors occurred during the admission and within 30 days before the onset of E. coli BSI. Clinical staging of pancreatic cancer was performed according to the American Cancer Society Tumor Staging Manual, 8th edition. TNM staging was performed first, with stages I and II being early and stages III and IV being late. e drainage tube was placed during the hospitalization but prior to E. coli BSI. Stayed in ICU was only this hospitalization before E. coli BSI. e number of admissions was restricted to admissions since the diagnosis of pancreatic cancer prior to E. coli BSI. e laboratory values were baseline values, which were the first value of the admission. e 90-day mortality rate (referred to the ratio of death within 90 days after the first positive blood culture) of the infected group and the control group was compared.

Ethics Statement.
is study was approved by the Research Ethics Committee of TMUCIH. Informed consent was obtained from all the patients according to the regulation of the Institutional Review Boards of TMUCIH in agreement with the Declaration of Helsinki.

Definition.
Hospital-acquired BSI was defined as the first positive culture obtained at 48 h after hospital admission or 48 h of discharge, along with clinical signs of active infection. When there was an unexplained fever and suspected to be a BSI, doctors would send at least one set of blood cultures and promptly sent for inspection. Blood samples (8-10 mL) were collected and auto-cultured by BACTEC 9050, 9120, or FX400 (Becton Dickinson, Franklin Lakes, NJ, USA) for 5 days. Positive samples were subcultured on blood agar (Jin Zhang Ke Ji, Tianjin, China) at 35°C for 24-48 hours depending on the results of Gram staining. Species identification and bacterial susceptibility tests were performed on a VITEK2 Compact automatic microbiological analysis system (bio-Merieux SA, Marcy l'Etoile, France); all coincidence rates were above 95%.

Statistical Analysis.
Bacterial resistance analysis, pathogen, and tumor-type distribution were performed by Whonet 5.6 software. SPSS 17.0 was adopted to perform univariate and multivariate analyses. Data of categorical variables were compared by Fisher's exact test; P < 0.1 was defined as a potential risk factor. e potential risk factors were included in the multivariate logistic regression model to analyze the independent risk factors; P < 0.05 indicates statistical significance; all tests were two-tailed. e results showed that hospitalization days ≥7 days, number of admissions ≥2 times, surgery, chemotherapy, the type of antibiotics used ≥2 species, albumin <40.0 g/L, and prealbumin <0.2 g/L were the potential risk factors for pancreatic cancer patients with E. coli BSI (P < 0.1); gender, age, clinical stage, intraoperative bleeding, radiotherapy, and other characteristics were not the potential risk factors for pancreatic cancer patients with E. coli BSI (P < 0.1). e 90-day mortality rate of the infected group was significantly higher than that of the control group (41.9% versus 8.6%, P < 0.001), see Table 3.

Analysis of Drug Resistance of Pancreatic Cancer Patients
Complicated with E. coli BSI.
ere was no significant difference in the resistance rates of antibiotics between E. coli BSI of pancreatic cancer and nonpancreatic patients. e ratio of E. coli producing extended-spectrum ß-lactamase

Discussion
BSI refers to the invasion of various pathogenic microorganisms (bacteria or fungi) into the bloodstream and is a serious systemic infectious disease [14]. e proportion of Gram-positive BSI has increased in the last 20 years, but Gram-negative BSI still accounts for about 50%, and E. coli is the leading cause [15,16]. Patients with malignant tumors often require surgery, high-dose chemoradiotherapy, antibiotics, and various invasive procedures; the malignancy itself tends to increase the chance of E. coli infection [17]. Blood culture is the gold standard for the diagnosis of E. coli BSI [18], but its diagnosis takes a long time, which will delay clinical and timely treatment. Besides, the distribution of pathogens and the risk factors for BSI are different due to different diseases and tumor types [19]. Pancreatic cancer has a high degree of malignancy, a low surgical resection rate, and a poor prognosis [20]. e occurrence of BSI may further increase the difficulty of treatment. erefore, clinical analysis of the risk factors for E. coli BSI in patients with pancreatic cancer is essential for preventing the potentially high-risk populations and the timely, effective treatment of patients.
We analyzed the baseline characteristics in the infected and control groups that may be related to infection [21][22][23], including gender, age, days of hospitalization, clinical stage, number of admissions, surgery, intraoperative bleeding, chemotherapy, radiotherapy, and other characteristics. Our study revealed that hospitalization days ≥7 days, number of  days ≥7 days, surgery, and chemotherapy were the independent risk factors for E. coli BSI. Pancreatic cancer patients have a long-term adjuvant and neoadjuvant treatment, radiation therapy, combined with various serious underlying diseases and multiple surgical treatments leading to prolonged hospital stay [24]; this gradually reduces the immune function, thus increasing the chance of E. coli BSI. A study has also shown that patients with more extended hospital stays are prone to BSI [25]. erefore, doctors and nurses in the hospital should strictly carry out the aseptic operation, i.e., regularly disinfect the medical equipment and related wards and departments, to reduce the incidence of BSI. Surgery is a kind of trauma to the human body, which will cause the patient's resistance to decline, causing BSI [26]. Besides, with the rapid development and popularization of medical technology, various implant operations such as drainage tubes, catheters, central venous catheters, etc., play not only a therapeutic role but also pose a risk to BSI; therefore, implantation was also an important risk factor [27]. Our data showed that surgery was an independent risk factor for E. coli BSI of pancreatic cancer patients.
In the results of this study, 67.7% of pancreatic cancer patients with E. coli BSI received chemotherapy (adjuvant or/and neoadjuvant treatment). Neoadjuvant treatment helps kill cancer cells and reduce the tumor implantation caused by surgery. Adjuvant therapy can effectively improve the treatment effect and reduce recurrence and metastasis. Adjuvant chemotherapy with gemcitabine is a standard care for resected pancreatic cancer [28], not only delayed recurrence but also improved survival compared with surgery alone [29]. However, while killing cancer cells, chemotherapy also kills normal immune cells. For instance, the dose-limiting toxicity of gemcitabine is myelosuppression, leading to decreased neutrophils and platelets, which reduces the body's immunity. Simultaneously, chemotherapeutic drugs such as doxorubicin can easily damage the intima of the blood vessels [30], leading to partial venous catheterization blockage, causing BSI. erefore, chemotherapy was an independent risk factor for E. coli BSI of pancreatic cancer patients. Cancer patients often have infections in other sites, and different types of antibiotics are often used for different infection sites [31]. We found that using two or more antibiotic types was a potential risk factor for E. coli BSI.
Neutrophils are the most abundant white blood cells, accounting for 50%-70% of the total white blood cells; they are the first barrier of the body's defense. When inflammation occurs, neutrophils will penetrate the vascular endothelium, and then enter the site of inflammation to exert a bactericidal effect. Patients with neutropenia may be prone  to bloodstream infections. However, we found that neutropenia is not directly related to E. coli BSI, which may be due to patients with pancreatic cancer suffering from chemoradiotherapy, which inhibits bone marrow function [24]. Meanwhile, they received leukocyte ascending therapy, resulting in a normal number of white blood cells and neutrophils, but they usually failed to exert anti-inflammatory effects. Albumin and prealbumin can be used as monitoring indicators of the nutrients in the body. When the nutritional status of the patients is poor, the immune function of the body decreases, and albumin and prealbumin decrease accordingly [32]. In our analysis, albumin <40.0 g/L and prealbumin <0.2 g/L were the potential risk factors, indicating that patients with pancreatic cancer were prone to E. coli BSI when their nutritional status is low.
In our study, the proportion of ESBL produced by E. coli was about 50%, which was lower than that reported in other studies [33]. e ceftriaxone resistance rate was significantly higher than that of ceftazidime, which may be related to the main cefotaxime (CTX) type of ESBL in this region. Carbapenems, cephamycin, and piperacillin/tazobactam can be used as the first choice for empirical use, but doctors should adjust the drug according to the drug susceptibility results of clinical microbiology to reduce the occurrence of superior and restricted antibiotic resistance.
However, this study also has some limitations. It is a single-center study, which can be combined for pancreatic cancer in North China, and the number of cases was small. e diagnosis and treatment of pancreatic cancer patients with E. coli BSI need further multi-center, large sample size research to promote its application.
In summary, pancreatic cancer patients with E. coli BSI exhibited higher mortality than control groups; patients with suspected E. coli BSI should be promptly drawn for blood culture. Hospitalization days ≥7 days, surgery, and chemotherapy are the independent risk factors for E. coli BSI; this allows us to identify patients at potential risk and perform preventive treatment in a short period. Early use of medication, while timely adjustment based on clinical drug sensitivity results, will also help reduce patients' mortality.
Data Availability e data that support the findings of this study are available from the archives of the TMUCIH, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request to the corresponding author (Yueguo Li) and with the permission of the TMUCIH.

Ethical Approval
Our study is a retrospective medical records data collection and analysis. All information was made anonymous before being made available for research. is study was approved by the Research Ethics Committee of TMUCIH.

Consent
Informed written consent was obtained from all the patients according to the regulation of the Institutional Review Boards of TMUCIH in agreement with the Declaration of Helsinki.

Disclosure
Changsen Bai and Xiuse Zhang are the two co-first authors.

Conflicts of Interest
e authors declare that they have no conflicts of interest.