Pneumococcal peritonitis: Still with us and likely to increase in importance

1University College Dublin, Dublin, Ireland; 2Department of Medicine, Faculty of Medicine and Dentistry; 3Department of Laboratory Medicine and Pathology, University of Alberta, Edmonton; 4Department of Paediatrics, University of Calgary, Calgary; 5EPICORE centre, University of Alberta, Edmonton, Alberta Correspondence: Dr Thomas J Marrie, Department of Medicine, Faculty of Medicine, Dalhousie University, Room C205, 5849 University Avenue, Halifax, Nova Scotia B3H 4H7. Telephone 902-494-1846, fax 902-494-7119, e-mail t.marrie@dal.ca Pneumococcal peritonitis is a relatively uncommon manifestation of Streptococcus pneumoniae infection (1), but nonetheless, a potentially life-threatening one (2) that requires prompt medical attention and treatment with antibiotics and, in some cases, surgery (3). Since the first descriptions of spontaneous bacterial peritonitis in English medical literature (in the United Kingdom in 1963 by Kerr et al [4] and in the United States in 1964 by Conn [5]), our knowledge of pneumococcal peritonitis has continued to evolve. Currently, this entity is classified into three groups: primary pneumococcal peritonitis associated with liver disease, including cirrhosis, viral hepatitis B or C, chronic renal failure and continuous ambulatory peritoneal dialysis or nephrotic syndrome; pneumococcal peritonitis sometimes presenting with an apparent genitourinary focus in young females; and secondary (or tertiary) peritonitis associated with gastrointestinal disease or after intra-abdominal surgery (1). All cases of invasive pneumococcal disease (IPD) in the province of Alberta during the five-year period from 2000 to 2004 were reviewed, and 23 cases of peritonitis due to S pneumoniae were found. The purpose of the present paper was to discuss our experience with these cases, and illustrate how the clinical picture of pneumococcal peritonitis has changed. In addition, patients with pneumococcal peritonitis were compared with those who had other manifestations of IPD to determine the differences.

BaCKGround: Pneumococcal peritonitis is uncommon and poorly understood.Methods: As part of a five-year study (2000 to 2004) of invasive pneumococcal disease (IPD) in Alberta, all cases of peritonitis due to Streptococcus pneumoniae were reviewed and compared with all other cases of IPD.results: Twenty-three of 1768 (1.3%) IPD patients were found to have peritonitis.Patients with peritonitis were more likely to have cirrhosis, hepatitis C, alcoholism and HIV/AIDS, than the remainder of the patients with IPD.The all-cause mortality did not differ between the two groups.Peritonitis was classified as primary in nine (39%) patients, secondary in 12 (52%) patients, and genitourinary in females, specifically, in two (9%) patients.Pneumococcal serotypes causing peritonitis were under-represented in current vaccines -17% among peritonitis patients versus 53% for the remainder of IPD patients for the 7-valent pneumococcal conjugate vaccine, and 56% versus 86% for the 23-valent pneumococcal polysaccharide vaccine.ConClusIons: Peritonitis represents a small subset of patients with IPD, but one that is likely to grow in importance given the increase in the number of patients with hepatitis C and HIV, and the reduced coverage of peritonitis serotypes in currently available vaccines.isolates to the National Centre for Streptococcus (NCS) located in Edmonton, Alberta, for capsular serotyping and antimicrobial resistance testing.To ensure that case ascertainment was as complete as possible during the study period, all reported IPD cases were identified from the NCS, the provincial health office and two active-surveillance research projects -the Community Acquired Pneumonia Study (CAPS) in Edmonton, and the Calgary S pneumoniae Epidemiology Research (CASPER) study in Calgary, Alberta.Data from these four sources were combined to form the complete data set.Cases found in more than one source were counted once only.

data collection and definition of pneumococcal peritonitis
Trained research nurses reviewed the clinical records of all patients with IPD and recorded the data using standardized case report forms.The data were entered into FileMaker Pro version 8.5 (FileMaker Inc, USA) and transferred to SPSS version 11 (SPSS Inc, USA) for analysis.
The subset of patients with pneumococcal peritonitis were defined as those with a clinical diagnosis of peritonitis (abdominal pain with guarding and tenderness to palpation), and S pneumoniae isolated from the blood and/or peritoneal fluid.The records were re-reviewed by one of the authors (TJM) to ensure that the case definition was met.
Peritonitis was classified as primary; with an apparent genitourinary focus in young females; and secondary, as described by Capdevila et al (1).

antimicrobial susceptibility assays
Antibiotic susceptibility was determined using reference broth microdilution methods as recommended by the Clinical and Laboratory Standards Institute guidelines (9).The following antimicrobial agents were assayed: penicillin, ceftriaxone, chloramphenicol, erythromycin, clindamycin, tetracycline, trimethoprim-sulfamethoxazole, levofloxacin and vancomycin.All antimicrobial agents were purchased from Sigma-Aldrich Canada Ltd.The minimum inhibitory concentrations used to interpret susceptibility for S pneumoniae were those that were current at the time that the assay was performed (9).

results
There were 1768 IPD patients during the period under surveillance, of which 23 (1.3%) had peritonitis.Table 1 provides a comparison of patients with peritonitis with those who had other manifestations of IPD.The analysis showed an over-representation of Aboriginal persons and, as expected, 39% of patients with peritonitis had cirrhosis compared with only 3% of the remaining IPD patients.Hepatitis C was the major cause of liver disease among the patients with peritonitis (22%) compared with the remaining IPD patients (6%), and there was a high rate of admission to surgery at 22% versus 2% for the remainder of the cases.Also, patients with HIV/AIDS were over-represented among the patients with peritonitis compared with the remaining IPD patients (13% versus 3%).The all-cause mortality rate of 25% among the peritonitis patients was not significantly higher than that of 13% for the remainder of the IPD patients.All vaccine serotypes had a much lower frequency in the peritonitis patients versus those with other manifestations of IPD, as shown in Table 1.
Table 2 details the clinical features for each patient.Only five isolates were from the peritoneal fluid.which in part, was due to the presence of ascites in only nine patients.While there were six patients with nephrotic syndrome in the entire study, only two had peritonitis.
The mean (± SD) length of stay of 13±12.8days for patients with peritonitis was not significantly longer than the 8±25 days for the other patients.
Fifteen isolates were available for susceptibility testing (eight were not submitted to our laboratory).Ten of the isolates were from blood, and five were from peritoneal fluid.All isolates assayed were susceptible to penicillin, ceftriaxone, chloramphenicol, tetracycline, vancomycin, clindamycin and levofloxacin.Three (20%) were resistant to erythromycin and two were of intermediate susceptibility to trimethoprim-sulfamethoxazole.In comparison, data from the nonperitonitis IPD patients revealed that 9.7% of the isolates were resistant to penicillin, 8.3% to erythromycin and 18.8% to trimethoprimsulfamethoxazole.
Table 2 shows selected data for all 23 patients.Four were children, two of whom had nephrotic syndrome as a predisposing cause.What is evident from the table is the complexity of the patients.Only the two patients classified as 'peritonitis in a young female' had no comorbidities.Nine patients had ascites and nine had pneumonia.Seven patients suffered from alcoholism.Pancreatic disease was also prominent in two patients with pancreatic cancer, and three with pancreatitis.Ten patients had altered mental status; four patients were afebrile.Nine patients had primary peritonitis, 12 had secondary peritonitis and two were classified as peritonitis in a young female.

dIsCussIon
Just over 1% of all our patients with IPD had pneumococcal peritonitis.In a study of the epidemiology of IPD in the United States, Robinson et al (10) reported 15,860 cases that occurred between 1995 and 1998.They did not indicate the number of cases of peritonitis in their study, nor if any occurred.It is noteworthy that in our overall study population, 15% of patients had alcoholism and 3.7% had cirrhosis.
When categorized according to the schema proposed by Capdevila et al (1), 52% of our patients had secondary peritonitis, 39% had primary peritonitis and 9% were young females who had peritonitis presumably due to an ascending infection through the genital tract (11).Hemsley and Eykyn (11), in a review of the literature from 1995 to 1998, noted only 27 cases of pneumococcal peritonitis in women who were otherwise healthy.In this group, 13 of 17 (76%) patients who had blood cultures performed had S pneumoniae bacteremia.Seven had pelvic inflammatory disease, six had appendicitis, six had intrauterine contraceptive devices and two occurred postpartum.Patient 10 had a pelvic abscess and no other co morbidities.Patient 1 was a two-year-old girl who fit the original description of primary pneumococcal peritonitis.
Nielsen et al (12) identified 18 cases of pneumococcal peritonitis over five years in two Danish counties.They proposed a new classification of pneumococcal peritonitis: children -33% of their patients fit this category; women of child-bearing age -44%; and older patients -22%.Applying this classification to our patients, 17% were children, 22% were women of childbearing age and 61% were older patients.
Three (13%) of the patients in our study had HIV/AIDS, all of whom were also hepatitis C positive.Because HIV/AIDS has become a chronic disease with the advent of effective antiretroviral therapy, pneumococcal peritonitis will be an increasingly frequent complication of HIV/AIDS further complicated by hepatitis C (13,14).Indeed, cirrhotic patients with HIV and spontaneous bacterial peritonitis have a higher rate of S pneumoniae infection as the etiological agent of the peritonitis compared with non-HIV-infected patients (14).The long-term prognosis for HIV/AIDS patients with pneumococcal peritonitis is poor (14).
Spontaneous bacterial peritonitis is also known to be a severe complication of nephrotic syndrome, and yet one that despite being characteristic in children, tends to be rare in adults (15)(16)(17).S pneumoniae is the most common causative agent of spontaneous bacterial peritonitis in patients with nephrotic syndrome (15).
In our study, one patient had rheumatoid arthritis and one patient had lupus, an autoimmune disorder that has been diagnosed in increased incidence in patients with pneumococcal peritonitis (18).Our patient with rheumatoid arthritis also had alcoholism and cirrhosis.
Seven (30%) of our patients had cirrhosis, pneumonia and alcoholism.In a cirrhotic rat model of pneumococcal pneumonia, early and late defects in killing S pneumoniae were identified (19).It is well known that alcoholism is a risk factor for acquisition of pneumonia and for severity of pneumonia (20)(21)(22).Thus, there is double jeopardy for some of these patients.In all likelihood, peritonitis results from the hematogenous spread of S pneumoniae in this subset of patients (1).
Some cases of pneumococcal peritonitis can be nosocomial and can be precipitated by gastrointestinal bleeding (23).Five of our patients had gastrointestinal bleeding.
The pneumococcal serotypes that caused peritonitis in our patients are under-represented in all currently marketed vaccines for S pneumoniae, indicating that such vaccines are unlikely to eliminate pneumococcal peritonitis.
While the isolates that were available for antimicrobial susceptibility testing from the peritonitis patients were all susceptible to penicillin, the missing isolates limit the conclusion that we can draw from this observation.
We conclude that pneumococcal peritonitis is a small, but distinct subset of patients with IPD, and one that is likely to grow in importance given the increase in the number of patients with hepatitis C and HIV.
aCKnoWledGeMents: The authors thank the staff of the Acute Diagnostic Microbiology Laboratories in Alberta who submitted isolates from IPD cases to the National Centre for Streptococcus, Edmonton, Alberta.They acknowledge the contributions of Carol Mangan RN BScN, and the Data Collection Team who include Anne Witschen RN, Lynne Korobanik RN, Freda Anderson RN, Loy Bacon RN, Shannon Pyra RN, Janine Schouten BSc, Ambreen Mithani (nursing student) and Natalie Chui (University of Alberta [Edmonton, Alberta] student).The authors also thank Linda Hastie RN, the study research coordinator; Stephanie Hui BSc, research assistant; Janice Pitchko BSc MSc, research epidemiologist for designing and maintaining the study databases; and Heather Mangan BSc for her assistance with the database.
ConflICts of Interest: Drs Tyrrell, Kellner and Marrie received a grant-in-aid from Wyeth {Pfizer Inc, USA} for the present study.In addition, they have served on the advisory board relating to pneumococcal vaccines for Wyeth.referenCes

Table 2 Clinical features and outcome of peritonitis cases (n=23)
Data from SPSS version 11.0 (SPSS Inc, USA).F Female; GI Gastrointestinal; M Male; PF Peritoneal fluid; SPB Spontaneous bacterial peritonitis