Proton Pump Inhibitors Are Associated With Accelerated Development of Cirrhosis, Hepatic Decompensation and Hepatocellular Carcinoma in Noncirrhotic Patients With Chronic Hepatitis C Infection

Results From ERCHIVES

D. K. Li; P. Yan; A-B. Abou-Samra; R. T. Chung; A. A. Butt


Aliment Pharmacol Ther. 2018;47(2):246-258. 

In This Article


In this large, national, well-defined cohort of HCV-infected veterans without a prior diagnosis of cirrhosis, we found that proton pump inhibitor use is independently associated with an accelerated progression to cirrhosis independent of having attained sustained virologic response. Moreover, we have demonstrated a strong association between proton pump inhibitor exposure in this group and an increased risk of hepatic decompensation and hepatocellular carcinoma. To our knowledge, this is the first study to demonstrate an association between proton pump inhibitors and acceleration of hepatic fibrosis in the pre-cirrhotic population. Prior retrospective studies that have demonstrated an association between proton pump inhibitor use and various liver-related complications in cirrhotics have generally had small or geographically limited populations and have reported on a single outcome. Our study expands the current knowledge by being a large, national study of an unselected population and reporting on 3 major clinical outcomes of interest in these patients.

Proton pump inhibitors have recently been shown to substantially alter the composition of the gut microbiota in healthy individuals. Changes include a decrease in the diversity of the gut microbiome, a decrease in the bacterial families Lachnospiraceae and Ruminococcaceae, an increase in oral and upper GI tract bacteria including Streptococcaceae and Micrococcaceae and an increase in the bacterial class Gammaproteobacteria, which include Enterobacteriaceae. These changes are likely due to reduced acidity of the upper GI tract due to proton pump inhibitor use, allowing for the distal colonisation of microflora normally confined to the upper GI tract. Several recent studies have also investigated the composition of the gut microbiome in cirrhotics. Intriguingly, these showed a marked similarity to that of the proton pump inhibitor-treated microbiome, including a decrease in bacterial diversity, a decrease in "healthy" bacterial populations including Lachnospiraceae and an increase in pathogenic bacteria including Enterobacteriaceae and Streptococcaceae.[15–17] Moreover, increasing severity of liver disease was associated with more significant changes in the gut microbiome.

Dysbiosis is thought to contribute to the pathophysiology of hepatic fibrosis by leading to increased translocation of enteric bacteria, resulting in higher levels of bacterial products including endotoxin in the portal circulation, which subsequently incite an inflammatory reaction as they reach the liver leading to the development of fibrosis. This leads to the activation of Toll-like receptors, which recognise pathogen-associated molecular patterns.[25] Activation of Toll-like receptors on a variety of cell types, including Kupffer cells and hepatic stellate cells, subsequently leads to the production of pro-inflammatory cytokines (ie TNF-α, IL-6), which has been linked, with hepatic injury and to acceleration of hepatic fibrogenesis[26] as well as hepatocellular carcinogenesis.[27] Interestingly, variants of the Toll-like receptor 4 gene (TLR4) have been reported that modulate the risk for liver fibrosis in Caucasian patients with chronic HCV infection.[28,29] Translocation of enteric bacteria is thought to be driven by several factors, including bacterial overgrowth and increased intestinal permeability. The connection between proton pump inhibitors and small intestinal bacterial overgrowth has recently been evaluated in several studies and is thought to be due to the abrogation of the acidic host defence in the stomach.[30,31] Recently, proton pump inhibitor-induced bacterial overgrowth has been implicated in the pathogenesis of some disease states including nonsteroidal anti-inflammatory drug-induced enteropathy.[32] In one study, duodenal aspirates retrieved from patients with small intestinal bacterial overgrowth who had been treated with proton pump inhibitors were found to predominantly have overgrowth of Streptococcus, Enterococcus and members of the Enterobacteriaceae family, consistent with what is observed in the distal gut microbiome of healthy patients treated with proton pump inhibitors.[33] Although proton pump inhibitor use has not been directly linked with increased intestinal permeability to date, dysbiosis itself has been increasingly linked with chronic intestinal inflammation which leads impaired gut barrier function and bacterial translocation.[34] For example, repeated administration of the oral microbe Porphyromonas gingivalis leads to altered distal gut microbiota, downregulation of genes involved in maintaining intestinal wall integrity and higher amounts of bacterial DNA in the liver as well as serum endotoxin levels.[35] Overall, those with chronic liver disease and particularly those with cirrhosis may be particularly predisposed to the potentially harmful effects of proton pump inhibitors as this class of medication predominantly undergoes hepatic metabolism. Interestingly, the effect was only significant in individuals exposed to proton pump inhibitors and not to those exposed to histamine-2 receptor antagonists, which has been previously observed in a study investigating the association between bacterial infections in decompensated cirrhosis and acid suppression.[36] However, we cannot exclude the possibility that at extremely high exposures of histamine-2 receptor antagonists, there may exist a significant effect on the risk of developing cirrhosis. Possible rationales for this disparity may be the relative potency of these medications in reducing the acidity of the gastric milieu or may be related to other as of yet unappreciated effects of proton pump inhibitor exposure.

We hypothesise that proton pump inhibitors may accelerate the progression of liver disease and the development of complications of cirrhosis by further exacerbating the dysbiosis associated with liver disease itself. By promoting such a dysregulated gut microbiome, proton pump inhibitors positively feed into the destructive cycle of cirrhotic pathophysiology by further increasing levels of bacterial overgrowth and translocation which then leads to further inflammation and hepatic fibrosis. As proof-of-principle, a recent study demonstrated that increasing the percentage of intestinal Gram-negative endotoxin producing bacteria in the gut may accelerate the development of liver fibrosis.[37] Furthermore, while this manuscript was being prepared, data from another study were presented in which Atp4 (encoding the H+/K+-ATPase targeted by proton pump inhibitors) mutant mice with resultant achlorhydria were found to have exacerbated phenotypes of alcoholic liver injury and non-alcoholic steatohepatitis compared to wild-type mice.[38] Moreover, this was associated with overgrowth of Enterococcus species, which was sufficient by itself to cause worsened alcoholic liver injury.[38]

Given the potential deleterious impact of long-term proton pump inhibitor use in noncirrhotics with hepatitis C as highlighted in this study as well as in cirrhotics, the use of proton pump inhibitors in this group of patients needs to be carefully assessed. The indications for proton pump inhibitor use are relatively narrow and most commonly include the treatment of gastro-oesophageal reflux disease, peptic ulcer disease and dyspepsia. However, proton pump inhibitors represent one of the most frequently prescribed drugs in cirrhotics and are often prescribed inappropriately without clear indications.[6,39,40] Frequently, proton pump inhibitors are prescribed to cirrhotics as part of treatment for gastro-oesophageal varices. Although benefits of proton pump inhibitor treatment have clearly been demonstrated in the management of acute variceal bleeding and in the immediate post-endoscopic banding period to prevent rebleeding,[41,42] evidence for the use of proton pump inhibitors outside of this period is lacking. Proton pump inhibitors have been shown to be ineffective in the primary prevention of upper GI bleeds in cirrhotics, including those related to varices and portal hypertensive gastropathy.[43,44] Moreover, a recent systematic review regarding the use of proton pump inhibitors in the management of gastro-oesophageal varices concluded that the best available evidence only supports a 10-day course of proton pump inhibitors post-endoscopic variceal ligation reduce ulcer size.[45] Given the results of our study and the limited evidence to date for long-term use of proton pump inhibitors in the management of portal hypertension-related bleeding, proton pump inhibitor use in those with chronic liver disease as well as those with confirmed cirrhosis should be limited to situations where their benefits outweigh the risks.

Strengths of this study include the use of a large national cohort of US subjects, with long length of follow-up and serial measurements of clinical and laboratory parameters. Additionally, many potential sources of confounding were addressed including confounding by indication and temporal ambiguity by excluding patients with baseline cirrhosis, hepatocellular carcinoma and those who were exposed to proton pump inhibitors only after being diagnosed with cirrhosis. We also adjusted for various risk factors for advanced liver disease including alcohol use and sustained virologic response status and found that our observed associations remained robust to sensitivity analyses involving both of these risk factors. Moreover, to account for the possibility that patients with a higher risk of liver disease progression to cirrhosis and subsequent adverse hepatic outcomes may reflect a population that is at high risk of being prescribed a proton pump inhibitor, we adjusted for conditions that were associated with proton pump inhibitor use including gastro-oesophageal reflux disease, H. pylori infection and peptic ulcer disease in our final models. To further minimise selection bias, we included persons with indeterminate sustained virologic response status (ie persons with missing HCV RNA values to confirm sustained virologic response after entry into the cohort). The results of a parallel analysis excluding individuals of indeterminate status demonstrate similar findings to our primary analysis and can be found in the supplementary data section as Figures S1-S5 and Tables S6-S9.

Limitations of this study include its nonrandomised study design, which opens the study to potential selection bias and unmeasured confounding variables. We also note that proton pump inhibitor users had slightly higher body mass index as well as mildly increased rates of diabetes and alcohol use, all of which have been associated with increased severity of liver disease. However, after multivariate Cox regression analyses to adjust for these potential confounders, proton pump inhibitor use was still found to be an independent risk factor for adverse hepatic outcomes. Indeed, while residual confounding cannot be entirely excluded in such an observational study, with a very large cohort, we could control for many observed and well-described risk factors for cirrhosis and increased disease severity in our adjusted multivariable models. As such, our findings are unlikely to be due to confounding alone. Another limitation of this study was the use of a clinical score (the FIB-4 score) as a surrogate to identify those patients with cirrhosis as opposed to liver biopsy, transient elastography or radiologic findings. However, the FIB-4 is a validated and widely used marker of liver fibrosis progression[46] and has been frequently used in epidemiological studies as a surrogate for a diagnosis of cirrhosis.[47,48] Moreover, the use of such a large sample size may minimise the variance otherwise attributable to this index score. Another limitation of this study is its use of prescription data, which may not accurately reflect patient adherence and also does not account for any over the counter use of acid-suppressive therapy.

In conclusion, our study demonstrates a significant association between proton pump inhibitors and a higher risk of fibrosis progression, cirrhosis, hepatic decompensation and development of hepatocellular carcinoma among individuals with chronic HCV infection. This may be attributable to dysregulation of the gut microbiota leading to increased levels of pathogenic bacteria and bacterial translocation, resulting in increased hepatic inflammation and fibrosis. Although this study operates within the limitations of observational epidemiology and is not designed to determine causality, the findings here suggest that excessive and indiscriminate proton pump inhibitor exposure even in the noncirrhotic population with risk factors for chronic liver disease may lead to adverse hepatic events and accelerated progression of liver disease. As such, randomised prospective studies with histological and clinical end points as well as animal studies to clarify the effect of long-term proton pump inhibitor use on the pathophysiological mechanisms of hepatic fibrosis are very much needed to further address this question.