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

Disclosures

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

In This Article

Results

Baseline Clinical Characteristics of Study Cohort

We identified 53 651 individuals with confirmed HCV infection who were initiated on anti-HCV therapy during the study period. We excluded those with HIV coinfection (n = 898), positive HBsAg (n = 6564), baseline gastro-oesophageal varices (n = 1528), hepatic decompensation (n = 83), hepatocellular carcinoma (n = 540) or cirrhosis (n = 5606) as well as missing or undetectable HCV RNA at baseline (n = 4150) and missing baseline or follow-up laboratories to calculate FIB-4 score (n = 18 046) (Figure 1). Patients exposed to both proton pump inhibitors and histamine-2 receptor antagonists were excluded to remove potential confounding from dual exposure to acid-suppressive therapy (n = 4325). In addition, patients who only received proton pump inhibitors after the diagnosis of cirrhosis were excluded (n = 385) to assure the correct temporal relationship between proton pump inhibitor exposure and development of liver-associated outcomes.

Figure 1.

Cohort design flow diagram. HCC = hepatocellular carcinoma

After these exclusions, 5752 proton pump inhibitor users and 5774 proton pump inhibitor non-users were included in the primary analysis. A comparison of baseline demographic and clinical characteristics of these populations is shown in Table 1. The mean age of both groups was 53 years of age and both were predominantly male (95.7% vs 96.5%). Both groups had similar breakdown of race (61.7% vs 60.2% white) as well as median HCV RNA levels. Similar numbers of both groups completed a course of HCV treatment and the treatment regimens that were received were also similar between both groups. Furthermore, baseline aminotransferase levels, platelet count and baseline FIB-4 score (1.77 vs 1.77; P = .77) were comparable in both groups. The proton pump inhibitor-exposed group had somewhat higher rates of diabetes, obesity, history of alcohol abuse/dependence and statin use. Proton pump inhibitor users were less likely to achieve sustained virologic response compared to proton pump inhibitor non-users (51.94% vs 55.90%; P <.0001). In both groups, there was a similar number of individuals for whom sustained virologic response was unable to be determined due to lack of follow-up HCV RNA to confirm sustained virologic response (18.50% vs 17.72%; P = .28). Median length of follow-up was 93.4 months (IQR 62.8–125.9) among proton pump inhibitor users compared to 89.5 months (IQR 57.6–123.6) for non-users. Among proton pump inhibitor users, the median duration of exposure was 27.25 months (IQR 8–69).

Effect of Proton Pump Inhibitor Exposure on the Development of Cirrhosis

To assess the effect of proton pump inhibitor use on the risk of progression to cirrhosis, we performed a multivariate Cox regression analysis adjusting for univariate predictors of cirrhosis including body mass index, diabetes and alcohol use as well as baseline FIB-4 score. We also adjusted for the presence of indications for proton pump inhibitor use including gastro-oesophageal reflux disease, peptic ulcer disease and H. pylori infection to minimise confounding by indication. Proton pump inhibitor users were divided into groups based upon drug exposure: cDDD 30–180, cDDD 181–540, cDDD 541–900, cDDD >900 and an unexposed group. After adjusting for potential confounders, we found that proton pump inhibitor use was an independent risk factor for the development of cirrhosis (Table 2). For patients with relatively limited exposure to proton pump inhibitors, proton pump inhibitors did not significantly affect the risk of developing cirrhosis compared to non-users. However, with increasing proton pump inhibitor exposure, there is a gradual, dose-dependent increase in the risk of cirrhosis. For patients with cDDDs > 900, the adjusted hazard ratio of cirrhosis development was 1.32 (95% CI: 1.17, 1.49; P < .001), respectively, compared to non-users (Table 2). Furthermore, the cumulative probability of cirrhosis over 10 years of follow-up by Kaplan-Meier analysis was significantly higher in patients who received proton pump inhibitor therapy (P < .001 using the log-rank test; Figure 2) and that the risk of cirrhosis increased in a dose-dependent manner (P = .003 using the log-rank test; Figure 3). Other identified risk factors for cirrhosis included increased age, diabetes, alcohol abuse/dependence history and increased FIB-4 score. Statin use was associated with a decreased risk of cirrhosis (adjusted HR: 0.60; 95% CI: 0.54, 0.67; P < .001), consistent with previous studies.[18,21]

Figure 2.

Probability of cirrhosis development in patients with chronic hepatitis C stratified by proton pump inhibitor use. Cirrhosis was defined as FIB-4 score > 3.5. P values calculated by log-rank test. PPI = proton pump inhibitor; cDDD = cumulative defined daily dose

Figure 3.

Kaplan-Meier curve for probability of cirrhosis-free survival in patients with chronic hepatitis C stratified by exposure level to proton pump inhibitors. Cirrhosis was defined as FIB-4 score > 3.5. P values calculated by log-rank test. PPI = proton pump inhibitor; cDDD = cumulative defined daily dose

We next sought to address numerous sources of potential confounding in our study. Given that failure to achieve sustained virologic response is one of the most powerful predictors of cirrhosis development; we first assessed contributions of potential confounding by sustained virologic response status. To do so, we performed Kaplan-Meier analysis of cirrhosis-free survival among proton pump inhibitor users and non-users stratified by sustained virologic response status. In this analysis, we found that among the subgroup of individuals who achieved sustained virologic response, exposure to proton pump inhibitors remained significantly associated with accelerated development of cirrhosis (P = .03; Figure 4). This association was also significant among proton pump inhibitor non-users (P = .003).

Figure 4.

Kaplan-Meier curve for probability of cirrhosis-free survival in patients with chronic hepatitis C stratified by proton pump inhibitor use and SVR status. SVR = sustained virologic response

Alcohol use is another independent risk factor for the development of cirrhosis. As such, we conducted additional sensitivity analyses excluding those with a diagnosis of alcohol abuse/dependence at baseline or during follow-up. In this cohort, proton pump inhibitor exposure remained significantly associated with an increased risk of cirrhosis with an adjusted hazard ratio of 1.22 for cDDD > 900 (95% CI: 1.06, 1.4; P = .01; Table S1).

We also wanted to ensure that bias was not introduced into this study as a result of minimal exposure to proton pump inhibitors prior to a diagnosis of cirrhosis, as we hypothesise that low levels of exposure to a relatively nonhepatotoxic drug such as proton pump inhibitors are unlikely to significantly influence the course of a chronically evolving disease like cirrhosis. As such, we determined the level of drug exposure among proton pump inhibitor users prior to achieving a diagnosis of cirrhosis. We found that among proton pump inhibitor users who developed incident cirrhosis (n = 902), 67.3% and 31.9% were exposed to > 180 cDDDs and > 900 cDDDs prior to a diagnosis of cirrhosis respectively. The mean pre-cirrhotic drug exposure among proton pump inhibitor users was 874 cDDDs. Overall, among proton pump inhibitor users, 58.7% of proton pump inhibitors were prescribed prior to a diagnosis of cirrhosis. These findings show that a spurious association was unlikely to be introduced into our study because of trivial pre-cirrhotic exposure to proton pump inhibitors.

Finally, we sought to assess whether the effect that we observed on cirrhosis development was specific to proton pump inhibitors or was generalisable to acid-suppressive therapy in general. As such, we performed a parallel analysis using identical methods on individuals with exposure to histamine-2 receptor antagonists. The overall demographic and clinical characteristics of the cohort of histamine-2 receptor antagonist users (n = 886) vs non-users (n = 10 640) are described in Table S2. We found that histamine-2 receptor antagonist use was not significantly associated with an increased likelihood of developing cirrhosis (adjusted hazard ratio for cDDD > 900: 1.07; 95% CI: 0.73, 1.58; P = .73); Table S3).

Proton Pump Inhibitor Effect on the Development of Hepatic Decompensation

We also evaluated the association of proton pump inhibitor use in our cohort with advanced complications of chronic liver disease, such as hepatic decompensation. Using multivariate Cox regression analysis and adjusting for univariate predictors of cirrhosis and risk factors for proton pump inhibitor use, proton pump inhibitor use was shown to be independently associated with a significantly higher risk of hepatic decompensation with adjusted hazard ratios for cDDD 181–540, 541–900 and > 900 of 3.46 (95% CI: 2.09, 5.75; P <.001), 3.72 (95% CI: 2.06, 6.72; P < .001) and 3.79 (95% CI: 2.58, 5.57; P < .001) respectively (Table 3). Notably, histamine-2 receptor antagonist users did not exhibit the same increase in risk of hepatic decompensation regardless of dose (Table S4). Over 10 years of follow-up, we also found that proton pump inhibitor use with an accelerated risk of developing a decompensation event (log-rank test; P < .001; Figure 5) and the risk of hepatic decompensation increased in a dose-dependent manner (log-rank test; P = .002; Figure 6).

Figure 5.

Kaplan-Meier curve for probability of hepatic decompensation in patients stratified by proton pump inhibitor use. P values calculated by log-rank test. PPI = proton pump inhibitor; cDDD = cumulative defined daily dose

Figure 6.

Kaplan-Meier curve for probability of hepatic decompensation in patients stratified by exposure levels to proton pump inhibitors. P values calculated by log-rank test. PPI = proton pump inhibitor; cDDD = cumulative defined daily dose

Proton pump inhibitor use was also associated with a significantly increased risk of developing hepatocellular carcinoma with a confounder-adjusted hazard ratio for cDDD > 900 of 2.01 (95% CI: 1.5, 2.7; P < .001) (Table 4). In contrast, histamine-2 receptor antagonist use was not associated with an increased risk of hepatocellular carcinoma (Table S5). Other identified risk factors for hepatocellular carcinoma include increased age while statin use and achievement of sustained virologic response were associated with a decreased risk of hepatocellular carcinoma.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....