Goel GA, Deshpande A, Lopez R, Hall GS, van Duin D, Carey WD. Increased rate of spontaneous bacterial peritonitis among cirrhotic patients receiving pharmacologic acid suppression. Clin Gastroenterol Hepatol 2012;10:422-427. (Reprinted with permission.)
BACKGROUND & AIMS: Patients with cirrhosis frequently receive proton pump inhibitor (PPI) or H2-receptor antagonist therapies. We investigated whether acid-suppressive therapy is associated with spontaneous bacterial peritonitis (SBP) in cirrhotic patients with ascites. METHODS: We compared data from 65 hospitalized cirrhotic patients with paracentesis-proven SBP, collected from 2006 to 2009, with those of 65 contemporaneous, hospitalized cirrhotic patients without SBP (controls). We evaluated PPI use and analyzed the effects of covariates. RESULTS: Patients with SBP had a significantly higher incidence of recent (past 7 days) PPI use (71%) than controls (42%). Of patients with SBP, 68% had no documented indication for PPI therapy. Based on multivariable logistic regression analysis, subjects who had not taken PPIs in the past 90 days were almost 70% less likely to develop SBP than those who had taken PPIs in the previous 7 days. Subjects who took PPIs within 8 to 90 days before hospitalization were 79% less likely to develop SBP than those who took PPIs within 7 days before hospitalization. There was no significant difference between patients who received no PPI therapy in the previous 90 days versus those who had taken PPIs in the previous 8 to 90 days (P = .58). Hyponatremia was associated significantly with SBP. There were no significant differences in length of hospital stay or 30-day survival for the SBP and control groups. CONCLUSIONS: Pharmacologic acid suppression is associated with SBP in patients with advanced cirrhosis. Prospective studies are needed to determine the mechanism of this association and to determine whether reduced use of PPIs and H2-receptor antagonists reduce the incidence of SBP.
In the study by Goel et al., the authors investigated the relationship between proton pump inhibitor (PPI) administration and the occurrence of spontaneous bacterial peritonitis (SBP), a topic with a tremendous potential impact in the clinical management of patients with cirrhosis.
PPIs are the third highest-selling in the pharmaceutical market in the United States, with $13.9 billion in sales,1 while SBP remains one of the principal causes of bacterial infection in cirrhosis. The risks involved with PPI consumption have ranged from a mild increased risk of spine and wrist fractures in postmenopausal women2 to a significant increased risk of Clostridium difficile infection3 as well as hospital and community-acquired pneumonia.4
In addition to the study by Goel et al., three other studies have investigated the risk of SBP in patients with cirrhosis taking PPIs, the results of which are controversial.5-7 In two of these studies, an increased in SBP incidence was shown, whereas this was not demonstrated in the study by Campbell et al. The design of all four of these studies was similar: (1) all were retrospective reviews of the medications taken by cirrhosis patients hospitalized in a single center; (2) all patients with documented PPI ingestion were considered PPI users; and (3) in the absence of data, patients were considered PPI nonusers.
The difficulties in the collection of data are shown clearly by the high number of medical records that had been invalidated and not included in the final analysis of the different studies. In the current study, a total of 1,309 patients were admitted between January 2006 and December 2009, but only 130 patients were eligible for final inclusion in the study. In accordance with the inclusion/exclusion criteria, almost half of the total number of patients were excluded either because they were transferred from other hospitals with inadequate records (601 patients) or they lacked a clear medication list (59 patients). Similar problems occurred in the study by Choi et al.,7 in which 176 patients were included but 42 patients were excluded because of an unclear history of acid-suppressive medications. The inclusion criteria described above may lead to an underestimation of patients taking PPIs because the administration of PPIs was not specifically investigated at the time of admission.
The second comment on the study is the decision of the investigators to include, opposite to all previous studies, patients receiving antibiotic treatment. The rational for this decision is unclear and deserves a special comment.
Practically, all patients that were admitted with the diagnosis of SBP were receiving antibiotics, representing an atypical SBP population. The authors stated that there was no diminution of SBP incidence in those patients receiving antibiotics and that such patients in fact had a higher prevalence of SBP. This finding is unexpected and is not reported in previous studies. Moreover, as the authors mentioned, patients receiving antibiotics have been excluded in all previous studies assessing the possible role of PPI in the development of SBP. This is due, most likely, to a clear reduction of intestinal bacterial overgrowth with antibiotic therapy.
Intestinal bacterial overgrowth in cirrhosis, assessed by aspiration and culture of small bowel contents or by the hydrogen breath test, has been reported in several publications to be more frequent in patients with cirrhosis than in healthy subjects.8, 9
The reason why PPI could increase the risk of SBP is unknown, but most of the authors hypothesize that PPI could increase intestinal bacterial overgrowth (IBO) leading to bacterial translocation and subsequently to SBP. However, conflicting results have been shown in several trials assessing an association between PPI use and IBO.10-12
A recent study in patients without cirrhosis showed that PPI therapy does not predispose patients to IBO.13 In this large study, the prevalence of IBO was measured by glucose hydrogen breath test (GHBT) in patients on PPI therapy compared with those not on PPI therapy. A total of 1,191 patients were included, 566 (48%) of whom were taking PPIs. The GHBT positivity was similar between patients using PPIs and those not using PPIs.
Finally, an unexpected finding in the study we are discussing was the relationship between the time of PPI administration and the SBP occurrence. The authors found that subjects who had not taken PPIs in the past 90 days were approximately 70% less likely to develop SBP than those who had taken PPIs in the previous 7 days and that there were no significant differences between patients who did not receive PPI therapy in the previous 90 days versus those who had taken PPIs in the previous 8-90 days. This suggests that only patients who took PPIs in the previous 7 days were at risk of developing SBP.
This unexpected finding has not been reported in previous studies, and due to the short period of PPI treatment, it is difficult to explain this finding within the context of an increase in IBO. Therefore, mechanisms other than IBO should be implicated in the increased risk of SBP in this and other studies.
In this regard, it has been suggested based in experimental data that acid-suppressive drugs may inhibit neutrophil functions and natural killer cell activity. However, the clinical significance of these findings is unknown.14
In conclusion, the role of PPI in the development of SBP is uncertain. The reason behind this uncertainty could be due, at least in part, to the retrospective nature of the studies and the difficulties to obtain reliable data from drugs that are available over the counter. Beyond the role of PPI in SBP occurrence, we should be concerned that around 50% of patients with cirrhosis are receiving PPIs without a firm indication.6 Prospective studies to evaluate the risk of SBP in patients with cirrhosis using PPIs are needed, but the design of those studies should be carefully planned.