Introduction

Spontaneous bacterial peritonitis (SBP) is one of the most serious complications in patients with cirrhosis. Among the causative pathogens, Escherichia coli is the most common Gram-negative bacillus [1, 2]. In the pathogenesis of SBP, bacterial translocation from the intestine into the mesenteric lymph nodes, followed by bacteremia and ascitic fluid inoculation due to impaired local and systemic immunity are considered to be two important steps [3, 4].

E. coli isolates causing SBP and bacteremia were genetically diverse and exhibited varying virulence profiles [5]. Phylogenetic group B2 was the most common [5, 6]. Outcome was not influenced by the phylogenetic group or the virulence profile [6]. Nosocomial SBP, severity of liver cirrhosis, and an antimicrobial profile with resistance to third-generation cephalosporins or extended-spectrum β-lactamases (ESBLs) were associated with poorer outcomes [2, 6, 7].

The K1 capsule of E. coli is believed to reduce the bacterial susceptibility to phagocytosis and complement the host defense system, and increase invasiveness [8, 9]. The K1 capsule antigen contributes to breach of the blood–brain barrier and subsequent neonatal meningitis and bacteremia [10–12]. Information is scarce about the outcomes of SBP caused by E. coli, with or without K1 capsular polysaccharide, in cirrhotic patients. Soriano et al. reported that the incidence of complications and mortality was similar in SBP patients infected with encapsulated E. coli strains with or without K1 [13].

The aims of this study were to examine and compare the prevalence and distribution of E. coli isolates among SBP and intestinal perforation (IP) with regard to K1 capsule antigen, phylogenetic grouping, and virulence factors. We further investigated the association between K1 capsule antigen and clinical manifestations and outcomes in E. coli SBP in cirrhotic patients.

Materials and methods

Patients and bacterial strains

From January 2004 to January 2012, E. coli isolates from peritoneal fluid were collected from patients at National Cheng Kung University Hospital in Taiwan. Data regarding patient characteristics, clinical manifestations, and in-hospital mortality were collected and analyzed. Only one isolate per patient was accepted. E. coli was obtained from aspirates of peritoneal fluid during surgery for IP or from ascites of cirrhotic patients.

SBP was defined by an ascitic fluid polymorphonuclear leukocyte count ≥250 cells/mm3 and a positive culture result in cirrhotic patients [2, 14]. The severity of liver cirrhosis was assessed according to the Child–Pugh score/classification and model for end-stage liver disease (MELD) score [15, 16]. IP was diagnosed by the surgical findings and/or imaging studies in patients with peritonitis. The causes of IP included trauma, ulcer, diverticulitis, infection (tuberculosis), ischemia, malignancy, or unknown.

Antimicrobial susceptibility

The susceptibility of E. coli strains, investigated using the disk diffusion method and interpretive criteria according to the Clinical and Laboratory Standards Institute (CLSI) guidelines 2011 [17], was determined for ampicillin, gentamicin, cefazolin, second-generation cephalosporins (cefuroxime, cefoxitin, or cefmetazole), third-generation cephalosporins (cefotaxime, ceftriaxone, ceftazidime, cefixime, or cefpodoxime), fourth-generation cephalosporins (cefepime or cefpirome), and fluoroquinolones (ciprofloxacin, levofloxacin, or lomefloxacin).

The E. coli strains from peritoneal fluid were identified using standard methods [18] and stored in 20 % glycerol at −70 °C until its use in all subsequent analyses.

Phylogenetic analysis and detection of virulence determinants

The phylogenetic grouping of the E. coli isolates was determined by a polymerase chain reaction (PCR)-based method, as previously described [19]. K1 capsule antigen and 14 uropathogenic virulence factor genes of E. coli were detected using PCR. Primer pairs specific for the K1 capsule gene, neuA, were K1-F: ATGATTACTCGACACTGTC; K1-R: AACAATCTCCGCTATTTCG. The size of the PCR products was 812 bp. Primer pairs specific for papG classes I to III, fimH, sfa, foc, afa, iha, hlyA, cnf1, iroN, iutA, ompT, and usp have been described previously [19, 20].

Statistical analysis

The Chi-square test or Fisher’s exact test (two-tailed) was used for the comparison of categorical variables, whereas the Wilcoxon rank-sum test was used for the comparison of continuous variables. A p-value <0.05 was considered to be statistically significant. All statistical analyses were performed using JMP software (SAS Institute Inc., Cary, NC, USA).

Results

A total of 54 and 41 E. coli strains derived from patients with SBP and IP, respectively, were included. Bacterial characteristics in relation to the source of E. coli strains are shown in Table 1. There were no significant differences in the distribution of phylogenetic groups and most of the virulence factors between the two groups. Phylogenetic groups D and B2 were the two most common groups in the SBP group. The prevalence of the K1 capsule antigen gene, neuA, was more common in SBP isolates compared to IP isolates.

Table 1 Comparison of the bacterial characteristics of Escherichia coli isolates derived from spontaneous bacterial peritonitis (SBP) and intestinal perforation (IP)
Full size table

The host characteristics of the 54 cirrhotic patients with SBP included 33 males (61 %) and had a mean age of 61 ± 12 years. The major causes of cirrhosis were hepatitis B virus (14, 26 %) and hepatitis C virus (17, 31 %). The median Child–Pugh score was 11 (range 6–14); the Child–Pugh classifications of A, B, and C were 1 (2 %), 9 (17 %), and 44 (81 %), respectively; and the median MELD score was 11 (range 7–22). There were no significant differences in age, gender, cause of cirrhosis, and presentations of SBP between patients in the K1 and non-K1 groups. The bacterial characteristics of the 54 E. coli isolates in relation to the K1 capsule antigen are shown in Table 2. Phylogenetic groups B2 and D were dominant in E. coli isolates with and without the K1 capsule antigen, respectively. The prevalence of virulence factors genes papG II, ompT, and usp was higher in E. coli isolates with the K1 capsule antigen. The antimicrobial resistance profile was similar between both groups.

Table 2 Bacterial characteristics of Escherichia coli isolates derived from cirrhotic patients with spontaneous bacterial peritonitis (SBP) in relation to the K1 capsule antigen
Full size table

Host and E. coli bacterial characteristics in relation to in-hospital death in cirrhotic patients with SBP are shown in Table 3. Twenty-six patients (48 %) died during hospitalization. There was a higher MELD score in the in-hospital death group (P = 0.0269), whereas there was no difference in the Child–Pugh score or classification between both groups. Presence of the E. coli K1 capsule antigen was significantly associated with in-hospital death, whereas a history of SBP in cirrhotic patients was inversely associated with in-hospital death.

Table 3 Host and Escherichia coli bacterial characteristics in relation to in-hospital mortality in cirrhotic patients with spontaneous bacterial peritonitis (SBP)
Full size table

Discussion

In patients with IP, E. coli bacteria escape the intestinal tract directly through a bowel perforation and enter the abdominal cavity. K1 capsular antigen predisposes to E. coli bacteremia in male infants with urinary tract infection (UTI) [12]. The pathogenic E. coli in cirrhotic patients with SBP is considered to be derived from bacterial translocation from intestinal flora [3, 4]. It is a pathogenesis different from that in IP. Our data showed a higher prevalence of the K1 capsule antigen in SBP compared to that in IP (28 % vs. 10 %, p = 0.0385), whereas there were no significant differences in phylogenetic grouping or other virulence factors. We speculate that the K1 capsule antigen is helpful in the escape from phagocytic clearance and complement activation in the lymphatic system and bloodstream following bacterial translocation. Therefore, the E. coli strain can reach the ascitic fluid as a result of bacteremia and then induce SBP.

The role of the K1 capsule antigen in different E. coli extraintestinal infections has not been well established. E. coli K1-related neonatal bacterial meningitis was associated with considerable mortality and morbidity [21, 22]. Male infants with E. coli bacteremic UTI have a higher prevalence of K1 capsular antigen than those with nonbacteremic UTI [12]. In a study of 137 adults with E. coli bacteremia, 16 (12 %) were caused by E. coli K1. There were no patient deaths from E. coli K1 bacteremia, while 16 of 48 patients with non-K1 bacteremia died [23]. Soriano et al. reported that encapsulated E. coli was identified in 27 of 37 cirrhotic patients with SBP, and they were associated with a higher complication rate than patients with nonencapsulated strains (93 % vs. 50 %, p < 0.01). However, the mortality rate was similar in SBP patients infected with encapsulated E. coli with or without K1 [13]. Our data showed that in-hospital death in cirrhotic patients with E. coli SBP was associated with a higher prevalence of the K1 capsule antigen. This result might suggest the role of the K1 capsule antigen in the outcome of cirrhotic patients with E. coli SBP.

papG II adhesin recognizes a specific receptor on uroepithelial cells [24]. Studies have shown that the papG II gene had a close association with E. coli upper UTI and bacteremia [20, 25]. ompT is a protease present in the outer membrane of E. coli; it confers resistance to urinary and antimicrobial cationic peptides [26, 27]. ompT and usp were frequently associated with UTI [19, 28], and a close association among ompT, usp, and kpsMT genes may be explained by co-selection and is beneficial for pathogenicity in UTI [28], whereas data on the relationships among K1 capsule antigen, papG II, ompT, and usp in E. coli infections is scarce. This study showed that E. coli strains with K1 capsule antigen were associated with a higher prevalence of papG II, ompT, and usp.

Our data showed that there were no associations between in-hospital death and most of the host factors, E. coli phylogenetic grouping, or other virulence factors, except the MELD score and K1 capsule antigen. Patients with a history of SBP were associated with less in-hospital death. It may be related to patient characteristics, in that those who survived the previous SBP may have earlier awareness of illness or better host immunity, nutrition status, or response to antimicrobial treatment. These may contribute to a better outcome in the following SBP episode. However, this is a single-center retrospective study with a small sample size. The small population analyzed may limit the statistical power of the results and their extrapolability.

In conclusion, we find a higher prevalence of the K1 capsule antigen in E. coli from SBP compared to E. coli from peritonitis caused by IP. We demonstrate a significant association between the K1 capsule antigen and in-hospital mortality in cirrhotic patients with E. coli SBP, and E. coli isolates with the K1 capsule antigen were more virulent and dominated by phylogenetic group B2. This observation suggests that the K1 capsule antigen may contribute to the development of E. coli SBP and poor outcome in cirrhotic patients.