Peritoneal dialysis–associated peritonitis contributes significantly to morbidity and mortality among peritoneal dialysis (PD) patients and to the utilization of health care. In an analysis of the Australian and New Zealand Dialysis and Transplant (ANZDATA) registry of 6639 incident PD patients, the 30-day mortality rate was 10% among patients with peritonitis. Peritonitis is a significant risk factor for PD technique failure with rates up to 50% for death censored technique failure in countries across the world. In this post, we discuss some recent high quality data from the Peritoneal Dialysis Outcomes and Practice Patterns Study (PDOPPS) and also add context from the recent International Society for Peritoneal Dialysis (ISPD) peritonitis guideline update which was partly informed by this study.
The key aspect of managing PD peritonitis involves timely diagnosis by analyzing the PD effluent with cell counts and cultures along with prompt initiation of antibiotics. Understanding the various factors that determine the outcomes of peritonitis is helpful in predicting clinical outcomes. To this effect, Al Sahlawi et al analyzed the data from PDOPPS and studied 1631 peritonitis episodes across seven countries. In this prospective, observational cohort study, the authors described regional differences in the cure rates of peritonitis across the involved countries and explored the impact of clinical, microbiological, and treatment characteristics on the likelihood of a cure following a peritonitis episode. “Cure” was defined as the absence of a subsequent peritonitis event, PD catheter removal, hemodialysis (HD) transfer, or death during the 50 days following onset of a peritonitis episode. The 50-day threshold was used in accordance with ISPD guidelines of defining relapsing and recurrent peritonitis.
Using a mixed effects logistic regression model, it was shown that peritonitis secondary to gram negative bacteria, polymicrobial and fungal etiology, use of continuous ambulatory PD (CAPD), and use of icodextrin were associated with lower odds of cure.
With regards to the microbiological organisms, the presence of gram negative bacteria (GNB) has been associated with worse outcomes including catheter loss and hospitalization in studies across the world and typically requires longer treatment, for 3-4 weeks depending on the organism. The mechanism of these worse outcomes likely includes difficulty in ensuring proper and timely antibiotic coverage in this setting and the association of GNB peritonitis with gut issues, such as perforations. Fungal peritonitis is another well-known indicator for PDantibiotics catheter removal as per ISPD guidelines with a 1C recommendation, and was associated with the highest odds of all the components of cure in the current study. In this context, the adage, “an ounce of prevention is better than a pound of cure” is a key lesson. As the ISPD guidelines also reiterate, it’s important to use an antifungal prophylaxis (eg, nystatin or fluconazole) when a PD patient receives any antibiotic prescription.
Theoretically, automated PD (APD) may be associated with higher risk of treatment failure due to inadequate serum antibiotic levels achieved with the rapid exchanges on the cycler. Patients are usually on a cycler at night and fluid is drained directly into the bathroom in a large portion of these cases, so there is also a chance that they may not notice the change in PD effluent promptly, delaying diagnosis. Another concern with the use of APD is the shorter dwell times on the cycler causing inadequate exposure to antibiotics. ISPD guidelines recommend sufficient dwell time to facilitate drug absorption, especially if the antibiotic has time dependent killing (eg, a minimal dwell time of four hours for Vancomycin). However, in this study, APD was associated with a greater likelihood of cure, despite all the theoretical reasons it should be worse. In hindsight, one can speculate that CAPD might be used more often when there is higher residual kidney function (RKF) and this may have led to inadequate antibiotic coverage. Needless to say, these results highlight the need to confirm theory with empiric data—and to always consider RKF in the drug dosing calculus.
Peritonitis is associated with an increase in membrane permeability which decreases ultrafiltration and could lead to fluid overload. A previous randomized control trial showed that the use of icodextrin could potentially improve ultrafiltration in these cases, as it is poorly absorbed into the blood. Glucose-containing solutions may also potentially worsen hyperglycemia among diabetics due to enhanced absorption across a leaky membrane. However the present study could not demonstrate the benefit of icodextrin on the PD peritonitis outcome (PD cell count on day 3 and mortality on day 120).
In another interesting twist, there was no significant impact of patient characteristics on the odds of cure in the PDOPPS study. Factors contributing to treatment response in PD peritonitis have been well analyzed in the past, with risk prediciton models developed to evaluate patient factors. A study by Xiang Lui et al developed a model based on high density lipoprotein cholesterol levels, fibrinogen levels, intestinal obstruction, diabetes mellitus, history of HD, and duration on PD. It showed an area under receiver operating characteristic curve (ROC) of 0.80 (95% confidence interval, CI 0.74-0.86, p<0.0001) to predict treatment failure. Similar risk prediction scores were developed by Nochaiwong et al with factors including diabetes, systolic blood pressure <90 mm Hg, and dialysate leucocyte count >1000 /mcL and >100 /mcL on days 3-4 and day 5, respectively. However, these risk scores were single centre studies, and needed external validation before being accepted for routine use. As seen from the PDOPPS data, after controlling for modality, microbiology, and other factors, the true role of patient characteristics seems to be marginal.
In terms of treatment strategies for GNB peritonitis, use of aminoglycoside or ciprofloxacin was associated with higher odds of cure compared to ceftazidime. This may be related to the presence of multidrug resistant GNB which usually responds to aminoglycosides. However, there is no data available about the type of bacteria that were isolated in this study. A Cochrane review of PD associated peritonitis did not show superiority of one antibiotic regimen over the other.
RKF warrants distinct mention, as the authors in the current study showed higher risk of relapse and recurrence in patients with significant RKF. This was also shown in a retrospective study where presence of urinary creatinine clearance >5 ml/min was associated with higher odds of relapsing and recurrent peritonitis when compared to anuric patients. Higher doses of antibiotics are needed in the presence of RKF to maintain adequate concentration of the drug in the blood and peritoneal fluid. RKF is particularly relevant for antibiotics like cefazolin, vancomycin, and ceftazidime which show a time-dependent killing, in comparison to aminoglycosides which show a concentration dependent killing.
In summary, it’s sobering to note that death occurred in 6% of patients (ranging from 2% in Japan to 15% in Thailand) within 50 days of a peritonitis episode. We cannot fix a problem without having high quality data to study and understand it—and these PDOPPS data go a long way to reveal what we still don’t know about this important issue.
Title: Variation in Peritoneal Dialysis–Related Peritonitis Outcomes in the Peritoneal Dialysis Outcomes and Practice Patterns Study (PDOPPS)
Authors: Muthana Al Sahlawi, Junhui Zhao, Keith McCullough, Douglas S. Fuller, Neil Boudville, Yasuhiko Ito, Talerngsak Kanjanabuch, Sharon J. Nessim, Beth M. Piraino, Ronald L. Pisoni, Isaac Teitelbaum, Graham Woodrow, Hideki Kawanishi, David W. Johnson, Jeffrey Perl