Culture-Negative Peritonitis: A Fifteen-Year Review

Abstract

Peritonitis is a serious complication in peritoneal dialysis (PD) patients; however, the clinical outcome of culture-negative peritonitis (CNP) is controversial. This retrospective study of CNP attempts to resolve this controversy. In 813 episodes of peritonitis, 202 episodes of CNP in 152 PD patients were reviewed. Two different methods of effluent culture were utilized during the study period. The incidence of CNP was lower with 50 ml centrifugation culture than 10 ml direct inoculation culture (20.7% vs. 35.7%; p < 0.05). The overall cure rate of CNP was greater than 80% of patients receiving cefamezine and gentamicin as initial therapy. Relapse within 30 days after completion of treatment happened in 9.6% of cases, and antibiotic therapy failed in 8.1% of CNP cases. In comparison with cured patients, patients with relapse or treatment failure are older (62.0 ± 15.0 years vs. 54.3 ± 15.3 years; p = 0.007) and have a higher rate of abdominal pain (91.4% vs. 69.3%; p = 0.007) and greater need for salvage therapy (54.3 % vs. 11.0%; p < 0.001). A history of antibiotic use or peritonitis within 30 days before onset of CNP was noted in 23.3% and 12.5% of cases, respectively, but was not associated with clinical outcome. The clinical outcome of CNP was benign in this study. Older age, abdominal pain, and need for salvage therapy were associated with an increased risk for relapse and treatment failure in CNP cases. Moreover, 50 ml centrifugation culture method was better than 10 ml direct inoculation culture in reducing the incidence of CNP.

Keywords:

• culture-negative peritonitis
• peritoneal dialysis

INTRODUCTION

Peritonitis is an important cause of morbidity and mortality in peritoneal dialysis (PD) patients.[1] Gram-positive organisms are the typical cause of PD-related peritonitis. However, a dialysate culture can be negative for a variety of clinical and technical reasons.[2–4] The guidelines of the International Society of Peritoneal Dialysis (ISPD) suggest that culture-negative peritonitis (CNP) should not account for more than 20% of peritonitis episodes.[5] However, in previous reports, the rate of CNP has been reported as 5–41%.[2–4],[6] The different incidence rates likely result from different definitions of peritonitis in these reports.

Clinical outcome for CNP is also controversial. Only two studies have analyzed clinical outcomes of CNP.[3],[4] Bunke et al. indicated that initial no-growth peritonitis had a benign outcome for half the catheter loss rate of culture-positive peritonitis.[3] On the other hand, Szeto et al. identified a primary response rate of 67.5% and a complete cure rate of only 37.7%. Early removal of peritoneal catheter in PD patients with CNP was recommended by the authors, especially for patients with recent history of peritonitis or antibiotic therapy.[4] Therefore, to evaluate the clinical outcome of CNP, a retrospective analysis of CNP in PD patients was performed.

PATIENTS AND METHODS

From January 1989 to April 2005, a total 902 patients with end-stage renal disease were treated with PD at Chang Gung Memorial Hospital. All CNP cases were reviewed retrospectively. The definition of CNP was as follows:

• abdominal pain or cloudy dialysate;

• initial leukocytosis in dialysate (white blood cell [WBC] count > 100/mL) with neutrophils > 50% (initial WBC < 100/mL with neutrophils > 50%, but WBC > 100/mL in following dialysate study was also included); and

• negative culture of the dialysate.

Patients with eosinophilic peritonitis (eosinophils > 10%) were excluded. All patients received follow-up for at least 30 days after completing antibiotic treatment. Demographic characteristics, underlying cause of renal failure, recent antibiotic therapy or peritonitis in 30 days, antibiotic regimen for peritonitis, and clinical outcome were reviewed. Two different methods of effluent culture were utilized during the study period. Prior to 1996, a well-mixed dialysate (10 ml) was inoculated into two blood culture bottles (5mL each) with aseptic technique. Identification of bacteria was performed using standard procedures. After 1996, sediment was cultured after centrifuging 50 ml dialysate (3000 × g for 15 minutes) before inoculation into to blood culture bottles.

The cure of peritonitis was defined as the complete resolution of peritonitis without relapse for 30 days following initial therapy (cefamezine and gentamicin) completion. Peritonitis relapse was defined as recurrence of peritonitis by any organism within 30 days after completing antibiotic therapy. The organism species resulting in peritonitis relapse were recorded. Treatment failure was defined as the required removal of the peritoneal catheter due to failed antibiotic therapy. When patients had persistent or progressive abdominal pain, fever, or cloudy dialysate under the initial antibiotic regimen, salvage antibiotics were prescribed by the individual physicians. Salvage antibiotics generally consisted of one or two of following antibiotics: vancomycin, third generation cephalosporin, amikacin, or ciprofloxacin.

Statistical Analysis

Statistical analyses were performed using STATVIEW software for Windows. Continuous variables are expressed as mean ± standard deviation (SD). Nominal variables are expressed as percentages. Factors associated with clinical outcome were analyzed using chi-square test or Students' t-test where appropriate. A value of p < 0.05 was considered statistically significant.

RESULTS

Demographics

Between 1989 and 2005, 813 peritonitis episodes occurred in our unit. After excluding cases of eosinophilic peritonitis (eight episodes), 202 episodes of CNP in 152 patients were analyzed. Of these, 116 patients experienced only one CNP episode, whereas 36 patients accounted for 86 episodes (26 patients, 2 episodes; 8 patients, 3 episodes; 1 patient, 4 episodes; and 1 patient, 6 episodes). Table 1 presents the demographic data for the 152 patients.

Table 1 Demographic data of 152 patients

Sex (M: F)	62: 90
Age (years)	53.6 ± 15.3
Duration on peritoneal dialysis (months)	24.3 ± 14.0
Causes of end stage renal disease
Chronic glomerulonephritis	20 (13.1%)
Diabetic nephropathy	34 (22.3%)
Hypertensive nephropathy	17 (11.2%)
Gouty nephropathy	4 (2.6%)
Systemic lupus erythematous	3 (1.9%)
Others/unknown	74 (48.7%)

Rate of Culture-Negative Peritonitis with Different Culture Methods

Before 1996, well-mixed effluent dialysate (10 ml) was inoculated into two blood culture bottles (5 ml each). Between 1989 and 1995, of 224 peritonitis episodes, 80 were CNP (35.7%). After 1996, the method that inoculated centrifuged 50 ml effluent into blood culture bottles was used. Of 589 episodes of peritonitis in 1996–2005, CNP accounted for 122 episodes (20.7%). There was a significant decrease in the incidence of CNP after changing culture method (p < 0.05).

Factors Affecting Choice of Initial Antibiotics Regimen and Outcome

For further analysis, four episodes of CNP were excluded as patients died from other causes besides infection. Of 198 CNP episodes, cefamezine and gentamicin were administered as the initial antibiotic regimen in 178 CNP episodes (89.9%). Vancomycin with gentamicin or ciproxin was prescribed for six (3.0 %) and nine (4.5 %) episodes, respectively. Age, abdominal pain, fever, initial WBC count in the dialysate, and diabetic status were not associated with the prescription of initial antibiotic regimen (data not shown). The history of recent antibiotic use was a significant factor for deciding initial antibiotic regimen for CNP. In CNP cases receiving cefamezine and gentamicin as initial regimen, 20.8% of cases had a past history of recent antibiotic therapy. In contrast, in CNP cases receiving other regimens, 45% of cases had a history of recent antibiotic use (p = 0.01). In patients with a history of recent antibiotic therapy, 41.3% was due to exit-site infection and 32.6% was due to previous peritonitis.

Clinical Outcome of All Patients with Culture-Negative Peritonitis

The overall cure rate was 82.3% after 14 days of initial antibiotic therapy. Relapse happened in 19 CNP cases (9.6%). Table 2 lists the culture results of peritonitis relapse. Failed antibiotic therapy and catheter removal were indicated in 16 cases of CNP (8.1%). One patient died due to peritonitis complicated with septic shock after salvage therapy and catheter removal. Table 3 presents the factors associated with clinical outcomes of PD patients with CNP. Patients with relapsing peritonitis were older than cured patients (62.8 ± 16.7 years vs. 54.3 ± 15.3 years; p = 0.02). In comparison with cured CNP patients, the percentage of abdominal pain was significantly higher in patients with relapsing peritonitis (89.4% vs. 69.3%; p = 0.026) and treatment failure (93.7% vs. 69.3 %; p = 0.016), respectively. The rate of recent peritonitis history was also higher in patients with relapsing peritonitis and treatment failure (15.8%, 12.5% vs. 6.1%); however, there was no statistically significant difference. More patients with relapsing peritonitis or treatment failure receiving salvage therapy than cured patients (31.5% vs. 11.0%, p = 0.022; 81.2% vs. 11.0%; p < 0.001, respectively). If one considered patients with relapsing peritonitis and treatment failure as a poor response group of patients, factors associated with clinical outcome were shown in Table 4. Patients with poor responses tended to be older than cured patients (62.0 ± 15.0 years vs. 54.3 ± 15.3 years; p = 0.007). The rate of abdominal pain was higher in patients with poor response than in cured patients (91.4% vs. 69.3%; p = 0.007). Patients with poor responses also received more salvage therapy (54.3% vs.11.0%; p < 0.001). Fever, rate of recent peritonitis, and choice of initial antibiotic regimen were not associated with clinical outcome in patients with poor response and cured patients (p = 0.089, 0.098, and 0.078, respectively).

Table 2 Pathogens of relapse peritonitis

Organism	Number of cases
Staphylococcus epidermidis	2
Streptococcus viridans	2
Coagulase-negative staphylococcus	3
E. coli	2
Enterococcus faecalis	2
Pseudomonas aeruginosa	1
TB	1
No growth	7

Table 3 Factors associated clinical outcome in all patients

	Cure (n = 163)	Relapse (n = 19)	Treatment failure (n = 16)
Age (years)	54.3 ± 15.3	62.8 ± 16.7	61.1 ± 16.7
Diabetes (%)	22.7	31.5	6.2
Abdominal pain (%)	69.3	89.4	93.7
Fever (%)	23.3	36.8	37.5
Salvage therapy (%)	11.0	31.5	81.2
Recent antibiotics (%)	22.7	21.0	31.2
Recent peritonitis (%)	6.1	15.8	12.5

*p < 0.05 vs. cured group.

Table 4 Factors associated with cured and poor response cases in CNP of PD patients

	Cure (n = 163)	Poor response(n = 35)
Age (years)	54.3 ± 15.3	62.0 ± 15.0
Diabetes (%)	22.7	20.0
Abdominal pain (%)	69.3	91.4
Fever (%)	23.3	37.1
Cefamezine and gentamicinas initial therapy	89.5	91.4
Salvage therapy (%)	11.0	54.3
Recent antibiotics (%)	22.7	25.7
Recent peritonitis (%)	6.1	14.3

*p < 0.05, cured peritonitis vs. poor response cases.

Clinical Outcome of Patients Receiving Cefamezine and Gentamicin as Initial Antibiotic Regimen

Of 198 CNP episodes, 178 (89.9%) were prescribed cefamezine and gentamicin as the initial antibiotic regimen. To standardize initial therapy, patients who received cefamezine and gentamicin were chosen for further analysis. Table 5 presents the factors associated with clinical outcome. Age, rate of abdominal pain, and rate of prescribing salvage therapy were significantly higher in patients with poor response than in cured patients. Although rate of recent peritonitis history was higher in patients with poor response than in cured patients; there was no statistically significant difference (12.5% vs. 4.1%, p = 0.06). Diabetic status, fever, and recent antibiotic use also did not differ between poor response group and cured group (all p > 0.05).

Table 5 Factors associated with clinical outcomes in patients receiving cefamezine and aminoglycoside

	Cure (n = 146)	Poor response(n = 32)
Age (years)	54.5 ± 14.9	61.2 ± 14.7^†
Diabetes (%)	21.0	21.8
Abdominal pain (%)	69.2	90.6^†
Fever (%)	23.3	37.5
Recent antibiotic (%)	19.8	25.0
Salvage therapy (%)	10.9	53.1^†
Recent peritonitis (%)	4.1	12.5

*Poor response: relapse peritonitis and treatment failure.

^†p < 0.05, cured group vs. poor response group.

DISCUSSION

This retrospective study reviewed and analyzed 202 episodes of CNP in 152 PD patients. During the study period, two different methods of effluent dialysate culture were used. This study clearly demonstrated that inoculation with centrifuged 50 ml effluent dialysate was superior to direct inoculation 10 ml effluent dialysate into blood culture bottles, as suggested by ISPD guidelines.[4] The overall cure rate of CNP was 82.3%, and patients with older age, with abdominal pain, or requiring salvage therapy during treatments had the highest risk of relapse or treatment failure in this study.

Several reasons may account for high CNP rate (> 20%) in this study. First, specimens with a WBC count < 100 ml and neutrophils > 50% were included in this study. In normal peritoneum, very few neutrophils exist. Therefore, even when the initial WBC count of the effluent dialysate is < 100/mL in patients with abdominal pain or cloudy dialysate, the proportion of neutrophils > 50% in the effluent dialysate is a strong evidence of peritonitis.[7] Sewell et al. utilized a similar definition for peritonitis and identified a 41% incidence rate of CNP.[2] Second, two different techniques for isolating microorganisms from infected effluent dialysate were used during this study. The incidence of CNP differed significantly between these two periods. Lye et al. reported that the Bactec culture method can improve the recovery rate of organisms when the isolation rate of organisms is low in the conventional 50 ml centrifugation technique.[6] Bactec bottles were costly and not available at our unit. When using blood-culture bottles system, we preferred 50 ml centrifuged method in isolating organisms from effluent dialysate of PD patients with peritonitis. The analytical results in this study also support ISPD guidelines that culturing sediment after centrifuging 50 ml of effluent is preferred when using blood-culture bottles.[5]

The majority of patients (89.9%) with CNP received cefamezine and gentamicin as the initial antibiotic therapy, whereas the other patients (10.1%) received alternative regimens. When choosing an alternative regimen as initial therapy for CNP, recent antibiotic use is an important consideration. Seven protocols were prescribed for treating CNP in Szeto's study, whereas three combinations were used in the study of Bunke et al.[2],[3] However, no significant difference in the rate of peritonitis resolution between various regimens has been reported. In this study, approximately 90% of patients received the same regimen; therefore, antibiotic therapy can be standardized when analyzing factors associated clinical outcomes of CNP.

The clinical outcomes of CNP in previous studies are controversial.[3],[4] Bunke et al. demonstrated that the resolution rate of peritonitis, incidence of hospitalization, shift to hemodialysis, and incidence of death, were similar in CNP and culture-positive groups.[3] However, Szeto et al. identified that clinical outcome for CNP was unsatisfactory.[4] In this study, the outcome of CNP is benign, with an overall cure rate of 82.3% after initial antibiotic therapy.

Several reasons may explain the discrepancy in these studies. First, the definition of clinical outcome was not uniform. In the study by Szeto et al., primary response was defined as resolution of abdominal pain, dialysate clearing, and effluent neutrophils count < 100/mL on day 10 with antibiotics alone.[4] Salvage treatment in this series was only applied after initial therapy for 10 days failed. Additionally, mean duration of salvage therapy was 2.75 ± 1.64 days. Therefore, a different principle of peritonitis treatment likely influences clinical outcomes. Second, the definitions for complete cure and relapse differed in previous studies. In this study, the definition of complete cure and relapse are similar to that in ISPD guidelines.[5] In contrast, Szeto et al. defined complete cure as complete resolution of peritonitis without peritoneal catheter removal, salvage antibiotic therapy, or relapse within 120 days. Peritonitis relapse was defined as early (within 30 days of completion of antibiotic therapy) or late relapse (at 30–120 days of completion of antibiotic therapy), by any organism.[4] These differences also affect study results for clinical outcome.

In this study, overall rate of cure, relapse, and failure was 82.3%, 9.6% and 8.1%, respectively. When considering relapse and treatment failure as poor response group, the analytical results demonstrated that patients with a poor response were older and had a higher rate of abdominal pain and greater need for salvage antibiotic therapy than cured patients. Previous antibiotics may alter local bacterial flora and provoke the development of resistant bacterial strains.[8] Antimicrobial agents retained in effluent may also result in poor recovery of microorganisms by culture.[2] Previous study suggested that recent peritonitis and antibiotic therapy were associated with poor treatment response.[2],[4] A history of recent antibiotic use was recorded in 23.3% of CNP patients in this study but did not play a significant role in clinical outcome. However, as this was a retrospective study and the liberal prescription of antibiotics was common in Taiwan, the true prevalence of recent antibiotic use was likely underestimated. These facts may explain the discrepancy between this study and other studies. The diabetic status, rate of fever, initial antibiotic regimen, and history of recent peritonitis were also not significantly different between the poor response group and the cured group.

In conclusion, the clinical outcome of CNP was benign in this study with an overall cure rate of > 80%. Older age, abdominal pain and need for salvage therapy were associated with increased risk for relapse and treatment failure in CNP cases. Moreover, when the blood culture bottle system is used, culturing sediment after centrifuging 50 ml effluent dialysate is superior to 10 ml well-mixed effluent inoculation.