Reply

Abstract

We thank Dr Khan and Dr Assaraf for their responses to our publication.1Tsai C.F. et al.Gastroenterology. 2017; 152 (134-141)Abstract Full Text Full Text PDF Scopus (113) Google Scholar Dr Khan suggested that an odds ratio (OR) of <3 is more likely explained by confounding factors. After adjusting for all potential confounding factors, however, we identified a dose-dependent risk of hepatic encephalopathy (HE) among cirrhotic patients using proton pump inhibitors (PPI); with an OR of 3.01 at cumulative defined daily dose of >365. In response to Dr Kahn’s comment on low OR findings, PPI use of higher dose or longer duration (cumulative defined daily dose > 365) meet his criteria of significance. Dr Kahn suggested that PPI use made for a “channel bias” in the present cohort. Certainly, compared with non-HE patients, our HE patients were at greater risk of various comorbidities, including gastrointestinal bleeding, intra-abdominal infection, and pneumonia. It should be kept in mind, however, that the National Health Insurance Bureau of Taiwan strictly regulates the use of PPI, such that all prescriptions require endoscopic documentation of peptic ulcer disease or reflux esophagitis. Additionally, PPI use is not even allowed in those with variceal gastrointestinal bleeding. Therefore, the “channel bias” in the current cohort should not be as significant as Dr Khan suggests. We agree that consumption of alcohol and hyponatremia may precipitate HE and may confound our results. This has been mentioned as a limitation of the present paper. Unfortunately, information on alcohol consumption and laboratory data are not available in most population-based insurance research databases. Bearing this in mind, future prospective studies should confirm our findings with the addition of data on alcohol consumption and laboratory data. We had hypothesized that PPI associated enteric dysbiosis contributes towards HE occurrence; however, Dr Kahn claimed that PPI use is inconsistent with small intestinal bacterial overgrowth (SIBO). In response to Dr Kahn, we provide several sources of evidence suggesting that there is gut dysbiosis after PPI use.2Bajaj J.S. Gut Microbes. 2014; 5: 397-403Crossref PubMed Scopus (108) Google Scholar Besides its potential association with SIBO, PPI use has been linked with bacterial translocation and altered gut microbiota.3Lewis S.J. et al.Aliment Pharmacol Ther. 1996; 10: 557-561Crossref PubMed Scopus (117) Google Scholar Regarding PPI use and specific microbial alterations, 4 weeks of PPI use can induce changes in bacteria taxa associated with gastrointestinal bacterial overgrowth (increased Micrococcaceae and Staphylococcaceae) and taxa with Clostridium difficile infection (increased Enterococcaceae and Streptococcaceae, decreased Clostridiales).4Freedberg D.E. et al.Gastroenterology. 2015; 149: 883-885.e9Abstract Full Text Full Text PDF PubMed Scopus (200) Google Scholar In a large prospective cohort, Jackson et al5Jackson M.A. et al.Gut. 2016; 65: 749-756Crossref PubMed Scopus (468) Google Scholar also demonstrated that PPI use was associated with an altered microbiota composition of lower diversity. A recent meta-analysis covering 11 studies (and including 3134 patients) identified a significant association between SIBO and PPI use, having a pooled OR of 2.282 (95% confidence interval, 1.238-4.205). The analysis further suggests that discrepancies in the SIBO–PPI association between studies may be owing to different methods of diagnosis (ie, duodenal or jejunal aspirate culture as opposed to a glucose hydrogen breathing test).6Lo W.K. et al.Clin Gastroenterol Hepatol. 2013; 11: 483-490Abstract Full Text Full Text PDF PubMed Scopus (223) Google Scholar Given the data on PPI use and the associated SIBO7Sugimoto H. et al.Drug Metabol Dispos. 2013; 41: 683-688Crossref PubMed Scopus (21) Google Scholar and dysbiosis,4Freedberg D.E. et al.Gastroenterology. 2015; 149: 883-885.e9Abstract Full Text Full Text PDF PubMed Scopus (200) Google Scholar, 5Jackson M.A. et al.Gut. 2016; 65: 749-756Crossref PubMed Scopus (468) Google Scholar we believe that PPI-related dysbiosis can be one of the main reasons contributing to HE among cirrhotic patients. PPIs may also contribute to HE occurrence through microbiota-independent mechanisms, as proposed by Assaraf et al. They suggested PPI use may cause drugs or toxic metabolites to accumulate in the brain by an interaction with P-glycoprotein at the blood–brain barrier.7Sugimoto H. et al.Drug Metabol Dispos. 2013; 41: 683-688Crossref PubMed Scopus (21) Google Scholar Although this mechanism is fascinating, the details remain unknown. We suggest that future well-organized animal and human studies are necessary to prove this claim. Proton Pump Inhibitor Administration Triggers Encephalopathy in Cirrhotic Patients by Modulating Blood–Brain Barrier Drug TransportGastroenterologyVol. 152Issue 8PreviewWe read with interest the article by Tsai et al1 in a recent issue of Gastroenterology, which highlights the influence of proton pump inhibitors (PPIs) use on encephalopathy occurrence in cirrhotic patients. This case control study performed in a large series of cirrhotic patients demonstrated a relationship between PPIs intake and the onset of a first hepatic encephalopathy (HE) episode, outlining a dose-dependent risk (odds ratios from 1.41 to 3.01). The authors explained the increased risk of HE by an alteration of gut microbiota under PPIs treatment, which equilibrium also seems to be disrupted in cirrhotic patients, especially during HE. Full-Text PDF Proton Pump Inhibitors and the Possible Development of Hepatic Encephalopathy in Cirrhotic Patients: True Association or Residual Confounding?GastroenterologyVol. 152Issue 8PreviewIn their recent study, Tsai et al1 evaluated a possible association between proton pump inhibitor (PPI) use and the development of hepatic encephalopathy (HE) in patients with cirrhosis. This is the second report of which we are aware that has examined this possible association. The first was a retrospective study by Dam et al2 that derived its population cohort from 3 separate clinical trials that were not designed to evaluate an association between PPI use and HE. They reported a hazard ratio of 1.36 (95% confidence interval, 1.01–1.84) after adjusting for multiple factors, all of which were significantly associated with HE, and also noted that only 66% of their patients had a valid indication for PPI use. Full-Text PDF