Visual Abstract
Keywords: CKD, clinical trial, drug nephrotoxicity, epidemiology and outcomes, gastrointestinal medications, randomized controlled trials
Abstract
Key Points
In this post hoc analysis of a randomized controlled trial, the proton pump inhibitor pantoprazole led to a faster rate of eGFR decline as compared with placebo.
Additional studies are needed to determine the effect of proton pump inhibitors on those at higher risk of adverse kidney outcomes.
Background
Observational studies have found an association between proton pump inhibitor use and worsening kidney function. It is unclear whether these associations are causal. We conducted post hoc analyses to determine the effect of pantoprazole on kidney function using data from the Cardiovascular Outcomes for People Using Anticoagulation Strategies (COMPASS) trial, a 17,598-participant randomized trial comparing pantoprazole (8791) with placebo (8807).
Methods
The primary outcome was the rate of eGFR change. Rate of eGFR change was based on the two eGFR measures available: the eGFR at randomization and at the open-label extension study that enrolled at trial conclusion. Secondary outcomes included incident CKD (defined by eGFR <60 ml/min per 1.73 m2 at open-label extension or case report forms) as well as AKI, acute nephritis, and nephrotic syndrome.
Results
Eight thousand nine hundred ninety-one of the 17,598 participants randomized to pantoprazole/placebo (51%) had eGFR recorded at baseline and open-label extension enrollment and were included in the rate of eGFR change population (mean age 67 [SD 8] years, 22% female, mean baseline eGFR 75 [SD 17.5] ml/min per 1.73 m2). The mean duration between randomization and open-label extension eGFR was 3.3 (SD 0.8) years. The placebo rate of eGFR change was −1.41 (SD 4.45) ml/min per 1.73 m2 per year. The pantoprazole rate of eGFR change was −1.64 (SD 4.47) ml/min per 1.73 m2 per year. In adjusted analyses, pantoprazole had a 0.27 ml/min per 1.73 m2 per year greater decline in eGFR (95% confidence interval [CI], 0.11 to 0.43). The odds ratio for the effect of pantoprazole on incident CKD was 1.11 (95% CI, 0.98 to 1.25) and on AKI was 0.89 (95% CI, 0.65 to 1.21). There were five nephrotic syndrome outcomes recorded and one event of acute nephritis.
Conclusions
In this post hoc analysis of the COMPASS trial, pantoprazole resulted in a statistically significant greater rate of eGFR decline as compared with placebo.
Clinical Trial registry name and registration number:
Rivaroxaban for the Prevention of Major Cardiovascular Events in Coronary or Peripheral Artery Disease (COMPASS), NCT01776424.
Introduction
Proton pump inhibitors (PPIs) are one of the most commonly prescribed medications.1,2 They are frequently used chronically without a documented indication.3–5 Before the Cardiovascular Outcomes for People Using Anticoagulation Strategies (COMPASS) trial, there had been no long-term randomized controlled trials (RCTs) of PPIs to inform their safety. Observational studies have found associations between PPI use and adverse outcomes, including pneumonia,6 dementia,7 cardiovascular events, and mortality.8 PPIs are a known cause of acute interstitial nephritis,9 and observational studies report associations with AKI,10 CKD, and kidney failure.11–13 Multiple potential mechanisms link PPIs to impaired kidney function (chronic interstitial nephritis, vascular calcification, oxidative stress, and alterations of the gut microbiome12); the causality of these associations remains uncertain because of limitations inherent to observational studies.14–16
The COMPASS trial compared the effect of rivaroxaban, with or without aspirin (ASA), with ASA alone on the composite of cardiovascular death, stroke, or myocardial infarction. In a partial factorial design, participants not already receiving a PPI were randomized to pantoprazole or placebo to assess upper gastrointestinal events. The primary efficacy and safety analyses of the PPI randomization in the COMPASS trial were previously reported.17,18 CKD was a prespecified safety outcome of interest, and the results did not demonstrate a higher risk of CKD in those randomized to pantoprazole (odds ratio [OR], 1.17; 95% confidence interval [CI], 0.94 to 1.45; P = 0.15). However, there were concerns these analyses were low power because of insensitive CKD event definitions that did not systematically use laboratory data and a low incidence of CKD events (181 CKD events [2.1%] for those randomized to pantoprazole and 158 CKD events [1.8%] for those randomized to placebo). The original analysis of the COMPASS trial did not exclude a modest but clinically important effect on kidney function. We undertook additional post hoc analyses to better understand the effect of pantoprazole on kidney function in the COMPASS trial.
Methods
COMPASS Trial Participants and Study Design
The COMPASS trial was an international, multicenter, blinded, placebo-controlled, randomized trial using a 3×2 partial factorial design. The trial design and primary results have previously been published.19,20 There was a 30-day run-in period where potentially eligible participants received rivaroxaban-matched placebo twice daily and ASA 100 mg once daily. Study pantoprazole/pantoprazole-matched placebo was not administered during the run-in period. Participants with at least 80% adherence to treatment during run-in were eligible for randomization.
Patients with chronic coronary artery disease and/or peripheral arterial disease were computer randomized in a 1:1:1 ratio to rivaroxaban 2.5 mg twice daily plus ASA 100 mg daily, rivaroxaban 5 mg twice daily plus ASA-matched placebo, or ASA 100 mg daily plus rivaroxaban-matched placebo twice daily. The subset of patients not receiving PPI at baseline were also computer randomized 1:1 to pantoprazole 40 mg daily or pantoprazole-matched placebo, stratified by center. The Population Health Research Institute randomization and drug management system assigned participants a unique number for study treatment. Participants, site personnel, sponsor personnel, and data analysts were blinded. In these post hoc analyses, we focus on the participants randomized to pantoprazole/placebo.
To be included, patients with coronary artery disease alone were required to be 65 years and older or, if younger than 65 years, were required to have evidence of atherosclerosis of at least two vascular beds or have two additional risk factors (current smoker, diabetes mellitus, kidney dysfunction with eGFR <60 ml/min per 1.73 m2, heart failure, or nonlacunar ischemic stroke ≥1 month ago). Patients were excluded if they had a high risk of bleeding, stroke within 1 month or history of hemorrhagic or lacunar stroke, severe heart failure with ejection fraction <30% or New York Heart Association class 3 or 4 symptoms, eGFR <15 ml/min per 1.73 m2, need for dual antiplatelet or anticoagulant or other non-ASA antiplatelet therapy, contraindication to rivaroxaban/ASA or study procedures, noncardiovascular disease with poor prognosis or that increases risk of adverse reaction to study interventions, treatment with drugs at high risk of interactions, and pregnancy or breast feeding.
The trial was conducted in accordance with Good Clinical Practice21 and the Declaration of Helsinki and was approved by relevant health authorities and research ethics boards in all participating countries. Participants provided written informed consent. The study sponsor was Bayer AG, and the trial steering committee was composed of Population Health Research Institute investigators, national study leaders, and Bayer representatives. The COMPASS trial is registered at ClinicalTrials.gov (NCT01776424). This is a post hoc analysis of the COMPASS trial and was not prespecified in the original COMPASS trial statistical analysis plan. The COMPASS trial was funded by Bayer.
From March 2013 to May 2016, a total of 27,395 participants were randomized to the rivaroxaban/ASA arms, 9797 of which were excluded from the pantoprazole/placebo arm of the study, primarily (98.8%) because of the need for continuous PPI. A total of 17,598 individuals from 580 centers in 33 countries were randomized to pantoprazole or placebo.
eGFR, CKD, and Other Kidney Outcome Assessments
All participants had Chronic Kidney Disease Epidemiology Collaboration creatinine 2009–derived eGFR recorded at baseline on the basis of the most recent creatinine measure from the local laboratory. After randomization, participants were seen at months one and six and then every 6 months for 3 years. The trial included the following prespecified kidney safety outcomes of interest: incident diagnosis of CKD, AKI (only if not related to cardiovascular event or bleeding), acute nephritis (not caused by infection), and nephrotic syndrome. Follow-up case report forms assessed for a new diagnosis of CKD but did not specify criteria and did not record serum creatinine or eGFR. Events were also recorded using hospitalization, drug discontinuation, death, and serious adverse event case report forms. At the conclusion of the trial period, eligible participants were approached to participate in an open-label extension study. Participants entering open-label extension had Chronic Kidney Disease Epidemiology Collaboration creatinine 2009–derived eGFR recorded on the basis of the most recent creatinine measure from the local laboratory. All creatinine measures were required to be performed within the year before randomization and open-label extension enrollment.
Outcomes
The primary outcome of our post hoc analyses was the rate of eGFR change (ml/min per 1.73 m2 per year) calculated as the difference between eGFR at baseline and the start of open-label extension divided by the duration between the two measures. The prespecified statistical analysis plan included a provision for examining the outcome of 40% eGFR decline should there be a statistically significant difference in the primary outcome.
The secondary outcomes included incident CKD on the basis of a hierarchical composite outcome defined by (1) eGFR <60 ml/min per 1.73 m2 at the start of the open-label extension, (2) CKD as the reason for exclusion from open-label extension (eGFR <15 ml/min per 1.73 m2 was excluded from open-label extension), and (3) a new diagnosis of CKD or death due to kidney failure recorded during trial follow-up. Additional secondary outcomes of AKI, acute nephritis, and nephrotic syndrome were based on events recorded during trial follow-up derived from Medical Dictionary for Regulatory Activities (MedDRA) coding on case report forms.
Analysis Populations
Participants who were not randomized to pantoprazole/placebo, which includes all those who were on a PPI at baseline, were excluded from our analyses. The primary outcome (rate of eGFR change) analysis population included all participants with available eGFR at baseline and open-label extension enrollment. The incident CKD analysis population included all participants with baseline eGFR ≥60 ml/min per 1.73 m2. The AKI, acute nephritis, and nephrotic syndrome analysis population included all participants randomized to pantoprazole/placebo.
Nine thousand two hundred eighteen of the 17,598 participants randomized to pantoprazole or placebo were enrolled in the open-label extension study. Participants were most commonly excluded from open-label extension because of participant refusal (n=3224, 38.4%), site not participating (n=1584, 18.9%), and death (n=1337, 16.0%) (Supplemental Table 1). An additional 227 participants had no open-label extension eGFR recorded and were excluded from the rate of eGFR change analysis. The rivaroxaban part of the trial was stopped early for efficacy after a mean follow-up of 23 months. The pantoprazole part of the trial continued as planned for a mean follow-up of 3 years. The mean duration between randomization and open-label extension entry eGFR was 3.3 years.
Statistical Analysis
Baseline characteristics are summarized by each analytical population. Continuous variables are summarized as means and standard deviations. Categorical variables are summarized as counts and percentages. For the primary analyses for all outcomes, participants are analyzed in keeping with the intention-to-treat principle. Because our objective was to assess safety, additional on-treatment analyses were conducted, where participants who permanently discontinued assigned treatment were excluded. We analyzed the effect of pantoprazole on the rate of eGFR change using a mixed-effects linear regression model. Provided there was a statistically significant effect of pantoprazole on the rate of eGFR change, the outcome of 40% decline in eGFR was to be examined using mixed-effects logistic regression. A sensitivity analysis was undertaken whereby the open-label extension eGFR was imputed as 15 ml/min per 1.73 m2 for those excluded from open-label extension because of kidney function. For this sensitivity analysis, the duration between randomization and open-label extension was imputed as the mean duration. The effect of pantoprazole on the secondary outcomes of CKD composite, AKI, acute nephritis, and nephrotic syndrome were analyzed using a mixed-effects logistic regression model. Partially adjusted models were adjusted for the stratification factors (center and the rivaroxaban/ASA arm randomization) in keeping with guideline recommendations.22,23 Fully adjusted models included stratification factors as well as age (as a continuous variable), baseline eGFR (as a continuous variable), diabetes, hypertension, race (Black relative to all other races), and sex. Centers were included as random intercepts. All other variables were included as fixed effects. Unadjusted, partially, and fully adjusted models are reported for all outcomes. Point estimates with 95% CIs are reported. Subgroup effects on the outcome of rate of eGFR change were explored by assessing the interaction between the randomized group and trial eligibility (coronary artery disease alone, coronary artery disease and peripheral artery disease, and peripheral artery disease alone), sex, age (<65, 65–74, or >74 years ), baseline eGFR (>90, 60–90, 30 to <60, or <30 ml/min per 1.73 m2), diabetes, hypertension, and geographic region (North America, South America, Western Europe, Eastern Europe, and Asia Pacific). There were no adjustments made for multiplicity. All analyses were performed in Stata version 15.
Results
The number of participants for each outcome is summarized in Figure 1. Baseline characteristics by analysis population are summarized in Table 1 and Supplemental Table 2. Baseline characteristics of those enrolled in the open-label extension and included in the eGFR change analysis and those who did not enroll in the open-label extension and were excluded from the eGFR change analysis are presented in Supplemental Table 2. Those randomized to pantoprazole had a lower rate of enrollment in the open-label extension than those randomized to placebo (50% of participants randomized to pantoprazole versus 52% of participants randomized to placebo; OR, 0.92; 95% CI, 0.86 to 0.97). The mean placebo group eGFR decreased from 75.2 (SD 17.6) to 70.6 ml/min per 1.73 m2 (SD 17.9), for a mean rate of eGFR change of −1.41 ml/min per 1.73 m2 per year (SD 4.45). Mean pantoprazole group eGFR decreased from 74.8 (SD 17.3) to 69.6 ml/min per 1.73 m2 (SD 18.1), for a mean rate of eGFR change of −1.64 ml/min per 1.73 m2 per year (SD 4.47). Distribution of the mean rate of eGFR change by assigned treatment is presented in Figure 2. In fully adjusted analyses, pantoprazole resulted in a faster rate of eGFR decline by 0.27 ml/min per 1.73 m2 per year (95% CI, 0.43 to 0.11) as compared with placebo (Table 2). The results of the sensitivity analysis imputing open-label extension eGFR as 15 ml/min per 1.73 m2 for those excluded from open-label extension because of kidney function (n=58) were consistent with the primary analysis results (Supplemental Table 4). The fully adjusted overall effects and fully adjusted effects by subgroup are presented in Figure 3 and Supplemental Table 5, respectively.
Figure 1.
Flow diagram illustrating the analysis population for each kidney outcome. (A) Mean rate of eGFR change. (B) CKD composite outcome. (C) Other kidney outcomes (AKI, acute nephritis, and nephrotic syndrome). ASA, aspirin; PPI, proton pump inhibitor; Riva, rivaroxaban.
Table 1.
Baseline characteristics of the different analysis populations by outcome: rate of eGFR change and CKD composite (eGFR <60 ml/min per 1.73 m2 at open-label extension enrollment, recorded as ineligible for open-label extension because of kidney dysfunction, or a CKD event or death because of kidney failure during trial follow-up)
| Characteristic | Rate of eGFR Change (n=8991) | CKD Composite (n=13,650) | ||
|---|---|---|---|---|
| Pantoprazole (n=4396) | Placebo (n=4595) | Pantoprazole (n=6836) | Placebo (n=6814) | |
| Age, yr | 67±8 | 67±8 | 67±8 | 67±8 |
| Female sex, no. (%) | 980 (22) | 977 (21) | 1293 (19) | 1270 (19) |
| Baseline eGFR, ml/min per 1.73 m2 | 75±17 | 75±18 | 82±13 | 82±13 |
| Baseline eGFR category, ml/min per 1.73 m 2 , no. (%) | ||||
| <30a | 24 (1) | 21 (1) | — | — |
| 30 to <60 | 884 (20) | 934 (20) | — | — |
| ≥60 | 3488 (79) | 3640 (79) | 6836 (100) | 6814 (100) |
| Race, no. (%) | ||||
| Asian | 519 (12) | 551 (12) | 1094 (16) | 1088 (16) |
| Black | 41 (1) | 60 (1) | 79 (1) | 80 (1) |
| Other | 1099 (25) | 1154 (25) | 1451 (21) | 1466 (22) |
| White | 2737 (62) | 2830 (62) | 4212 (62) | 4180 (61) |
| Region, no. (%) | ||||
| North America | 583 (13) | 635 (14) | 996 (15) | 962 (14) |
| South America | 1206 (27) | 1234 (27) | 1537 (23) | 1560 (23) |
| Western Europe, Israel, Australia, or South Africa | 1053 (24) | 1128 (25) | 1834 (27) | 1794 (26) |
| Eastern Europe | 1084 (25) | 1100 (24) | 1458 (21) | 1499 (22) |
| Asia-Pacific | 470 (11) | 498 (11) | 1011 (15) | 999 (15) |
| Diabetes, no. (%) | 1584 (36) | 1648 (36) | 2542 (37) | 2550 (37) |
| HTN, no. (%) | 3285 (75) | 3473 (76) | 5018 (73) | 5046 (74) |
| CAD, no. (%) | 3955 (90) | 4067 (89) | 6136 (90) | 6135 (90) |
| Peripheral arterial disease, no. (%) | 1217 (28) | 1326 (29) | 1831 (27) | 1830 (27) |
Plus-minus values are means±SD. CAD, coronary artery disease; HTN, hypertension.
eGFR<15 was an exclusion criteria at study and open-label extension enrollment.
Figure 2.
Distribution (percent of participants) of rate of change of eGFR (ml/min per 1.73 m2 per year, CKD-EPI) by treatment group (placebo versus pantoprazole). CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration.
Table 2.
Partially adjusteda and fully adjustedb intention-to-treat analysis results examining the effect of pantoprazole versus placebo on the outcomes of rate of eGFR decline, CKD composite outcome, 40% eGFR decline, and AKI
| Event | Placebo Mean Rate eGFR Change, ml/min per 1.73 m2 per Year±SD | Pantoprazole Mean Rate eGFR Change, ml/min per 1.73 m2 per Year±SD | Partially Adjusted Analysisa | Fully Adjusted Analysisb |
|---|---|---|---|---|
| Pantoprazole Effect Estimatec (95% CI), P Value | Pantoprazole Effect Estimatec (95% CI), P Value | |||
| Rate of eGFR change, ml/min per 1.73 m2 per year±SD (n=8991) | −1.41±4.45 | −1.64±4.47 | −0.26 (−0.44 to −0.09), P = 0.004 | −0.27 (−0.43 to −0.11), P = 0.001 |
| Event | Event Rate Placebo, No. (%) | Event Rate Pantoprazole, No. (%) | Partially Adjusted Analysisa | Fully Adjusted Analysisb |
|---|---|---|---|---|
| Odds Ratio (95% CI), P Value | Odds Ratio (95% CI), P Value | |||
| CKD composite (n=13,650) | 646 (10) | 706 (10) | 1.10 (0.98 to 1.23), P = 0.10 | 1.11 (0.98 to 1.25), P = 0.09 |
| 40% eGFR decline (n=8991) | 139 (3) | 136 (3) | 1.03 (0.81 to 1.32), P = 0.81 | 1.03 (0.81 to 1.32), P = 0.79 |
| AKI (N=17,598) | 92 (1) | 81 (1) | 0.88 (0.65 to 1.19), P = 0.40 | 0.89 (0.65 to 1.21), P = 0.44 |
CI, confidence interval.
Adjusted for center and rivaroxaban/aspirin arm randomization.
Adjusted for age, sex, baseline eGFR, hypertension, diabetes, race, center, and rivaroxaban/aspirin arm randomization.
Rate of eGFR change (ml/min per 1.73 m2 per year) and odds ratios for dichotomous outcomes.
Figure 3.
Mean rate of eGFR change (ml/min per 1.73 m2 per year) pantoprazole as compared with placebo by subgroup and overall population in fully adjusted analyses (adjusting for stratification factors of center and rivaroxaban/ASA randomization as well as age as a continuous variable, baseline eGFR as a continuous variable, diabetes, hypertension, race—Black race relative to all other races, and sex). CAD, coronary artery disease; eGFR, eGFR in ml/min per 1.73 m2; PAD, peripheral arterial disease.
There were a total of 1352 incident CKD composite events, comprising 1261 (93%) with open-label extension eGFR <60 ml/min per 1.73 m2, ten (1%) with open-label extension exclusion for impaired kidney function (eGFR <15 ml/min per 1.73 m2), one (0.1%) death due to kidney failure, and 80 (6%) with CKD recorded on case report forms during follow-up. There were 646 CKD composite events (10%) in the placebo group and 706 (10%) in the pantoprazole group (OR, 1.11; 95% CI, 0.98 to 1.25).
The effect of pantoprazole on the outcome of 40% eGFR decline from baseline was explored as prespecified given the statistically significant difference in the mean rate of eGFR change. There was a total of 275 events, 139 (3.0%) in the placebo group and 136 (3.1%) in the pantoprazole group (OR, 1.03; 95% CI, 0.81 to 1.32). There were 173 AKI events, 92 (1.0%) in the placebo group and 81 (0.9%) in the pantoprazole group (OR, 0.89; 95% CI, 0.65 to 1.21) (Table 2). There were four nephrotic syndrome outcomes in the pantoprazole group (0.05%) and one in the placebo group (0.01%). There was one event of acute nephritis in the pantoprazole group (0.01%) and none in the placebo group. Given the low event rate, additional analyses for these outcomes were not performed.
For all outcomes, the results of partially adjusted analyses, fully adjusted analyses, and completely unadjusted analyses were consistent (Table 2 and Supplemental Table 6). The results from the on-treatment analyses were consistent with the intention-to-treat analyses (Supplemental Table 7).
Discussion
In this post hoc analysis of the COMPASS RCT, eGFR declined 0.27 ml/min per 1.73 m2 per year more quickly in the pantoprazole group as compared with placebo. This represents an approximately 1.2-fold greater rate of decline with a consistent effect on CKD. Within the limitations of a large degree of missing data, these results suggest a negative effect of pantoprazole on kidney function.
Our results are broadly consistent with observational studies demonstrating an association between PPI use and adverse kidney outcomes.11–13 Although pantoprazole is deleterious on kidney function, our findings importantly inform the magnitude of the effect. For example, reassuringly for patients at low risk of progressive CKD, PPIs may result in only marginal loss of kidney function, while effectively treating gastroesophageal reflux, improving quality of life,24 and, in the case of Barrett esophagus, reducing all-cause mortality.25 Conversely, in patients without a strong indication for acid suppressive therapy26 and/or at high risk of kidney failure, where even small excess losses of kidney function could result in the need for dialysis more quickly or complications of CKD, PPIs may be undesirable. These results therefore enable shared decision making.
The assessment of this risk–benefit profile is uncertain because we cannot know whether the excess loss of eGFR is absolute (i.e., a 0.27 ml/min per 1.73 m2 per year greater loss with pantoprazole regardless of whether the eGFR was declining at 1.41 ml/min per year, as in the COMPASS placebo group, or 4 ml/min per year, as may be seen in a higher risk population) or relative to the baseline rate of eGFR loss (increasing the rate of loss of eGFR by 1.2-fold such that for individuals declining at 4 ml/min per 1.73 m2 per year, pantoprazole use increases this by a factor of 1.2–4.8 ml/min per 1.73 m2 per year). If the effect of pantoprazole is relative, an excess loss of 0.8 ml/min per 1.73 m2 per year is roughly equivalent to, but in the opposite direction of, the beneficial effect of renin-angiotensin system blockade.27 In the COMPASS trial, the subgroup of patients enrolled on the basis of peripheral artery disease had a more rapid rate of eGFR decline, and in this group, the effect of pantoprazole was also larger, suggestive of a relative effect. However, these subgroup analyses should be interpreted with caution and viewed as hypothesis generating. Additional studies are needed to clarify the effect of PPIs in patients with faster rates of eGFR decline.
Our analyses address many of the limitations of observational studies examining the effect of PPIs on kidney outcomes. The issues of residual confounding and other potential biases (indication bias, selection bias, information bias, etc.) left doubt as to whether the harms associated with PPIs were causal. Prior RCTs were small and had relatively short follow-up. The COMPASS trial presented a unique opportunity to study the effect of a PPI in a large number of patients, for a long duration in a randomized placebo controlled trial. We were able to study the rate of change of eGFR, a proposed surrogate outcome for kidney disease progression that accurately predicts the treatment effect on the traditional clinical outcome of doubling of serum creatinine and/or kidney failure (R2=0.97; 95% Bayesian credible interval, 0.78 to 0.99).27
Our study has several limitations. The COMPASS trial was not designed to assess these kidney outcomes; all our analyses presented here are post hoc and without adjustment for multiple testing and should be interpreted in this light. Follow-up eGFR was assessed in 51% of the trial population that enrolled in the open-label extension, and restricting our analyses to these patients results in a nonrandomized comparison that may introduce bias. Importantly, bias in the assessment of kidney function would be introduced primarily if there was bias in the enrollment in the open-label extension study and/or bias in the ascertainment of kidney function at entry into the extension. Neither is likely given the focus of the trial on cardiovascular and bleeding events and that enrollment into the extension was largely based on patient preference or centers not participating, which did not vary by treatment group (Supplemental Table 1). Supportive of this, we did not find evidence that included patients differed from those excluded from eGFR change analyses (Supplemental Table 3). In addition, it should be noted that our eGFR change analyses compare those enrolled in the open-label extension who were randomized to pantoprazole with those enrolled in the open-label extension who were randomized to placebo. It should be noted that 2% more participants in the placebo group enrolled in the open-label extension (OR, 0.92; 95% CI, 0.86 to 0.97), and this may represent a bias in our results that is unpredictable. Having said that, participants with poorer health are generally thought less likely to continue enrollment in studies, and if this is true in COMPASS, we likely underestimate the difference between the pantoprazole and placebo eGFR at follow-up, although any such bias is likely to be minimally impactful given the small differential in enrollment. Our analyses may have underestimated the effect of pantoprazole on eGFR decline because the mean study time on pantoprazole was 3 years while mean duration between baseline and open-label extension was 3.3 years. CKD outcome analyses involving a larger proportion of the trial population were congruent with the analyses of change in eGFR, in that they showed a trend toward worse kidney outcomes in those randomized to pantoprazole, albeit not meeting conventional statistical significance. Analyses on the outcome of 40% eGFR decline were limited by low event rates, and we were unable to exclude a clinically meaningful effect. The eGFR measurements in follow-up were also taken only once and may not be sustained. This is expected to result in greater random error in outcome ascertainment and reduced power and appears to bias the results to the null.28 Despite this, we still demonstrated a clear effect on change in eGFR over time. Pantoprazole worsening longer-term kidney outcomes, such as rate of eGFR change and incident CKD without a signal for higher risk of AKI, is consistent with prior literature.29 However, there are important limitations to our AKI findings. Analyses on the effect of pantoprazole on AKI were limited by low event rates. AKI was defined on the basis of events recorded during trial follow-up derived from MedDRA coding. The reliance on MedDRA coding likely led to missing cases of community-acquired AKI. There was no requirement for eGFR reporting, nor explicit use of consensus definitions, such as Kidney Disease Improving Global Outcomes30 criteria. The lack of a uniform definition of AKI likely resulted in greater imprecision and higher measurement error.
In conclusion, in this post hoc analysis of the COMPASS trial, pantoprazole use was associated with a greater rate of decline in eGFR as compared with placebo. While the clinical significance of the effect size is unclear, it is important to consider both the potential benefits and risks of PPI use. In patients with no indication, which appears to be many patients prescribed PPIs,3–5 or low probability of benefit and at high risk of kidney failure, chronic use of a PPI may be a poor option. In patients with clear benefits and low risk of kidney failure, chronic PPI use may be a reasonable option. In either case, these analyses from the COMPASS trial can be used to help inform shared decision making and monitoring strategies for patients prescribed PPIs. Additional work is needed to clarify the magnitude of effect PPIs have in a population at higher risk of adverse kidney outcomes. These findings should not lead to avoiding the use of PPIs when a strong clinical indication is present, such as Barrett esophagus, healing of peptic ulcer disease, and as part of a regimen eradicating Helicobacter pylori. Serial monitoring of kidney function for patients on PPIs may be an option to help prevent more severe kidney function decline.
Supplementary Material
Acknowledgments
The COMPASS trial was funded by Bayer.
Footnotes
See related editorial, “Proton Pump Inhibitors and CKD: The Evidence Builds,” on pages 833–834.
Disclosures
Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/E638.
Funding
None.
Author Contributions
Conceptualization: Lonnie Pyne, Andrew Smyth, Michael Walsh.
Formal analysis: Lonnie Pyne.
Funding acquisition: Jacqueline Bosch, John Eikelboom, Salim Yusuf.
Investigation: Jacqueline Bosch, John Eikelboom, Salim Yusuf.
Methodology: Jacqueline Bosch, John Eikelboom, Lonnie Pyne, Michael Walsh.
Project administration: Jacqueline Bosch, John Eikelboom.
Resources: Jacqueline Bosch, John Eikelboom, Salim Yusuf.
Supervision: Jacqueline Bosch, John Eikelboom, Michael Walsh.
Writing – original draft: Lonnie Pyne.
Writing – review & editing: Jacqueline Bosch, John Eikelboom, Paul Moayyedi, Amber O. Molnar, Eva Muehlhofer, Lonnie Pyne, Andrew Smyth, Michael Walsh, Salim Yusuf.
Data Sharing Statement
Bayer has set up a portal that allows users to submit proposals, which are then vetted by a committee and, if approved, data are made available for analysis.
Supplemental Material
This article contains the following supplemental material online at http://links.lww.com/JSN/E637.
Supplemental Table 1. Reason for long-term open-label extension (LTOLE) exclusion by treatment arm.
Supplemental Table 2. Baseline characteristics of the other kidney outcomes (AKI, nephrotic syndrome, acute nephritis) analysis population.
Supplemental Table 3. Baseline characteristics of those enrolled in the open-label extension study and included in the eGFR change analysis and baseline characteristics of those who did not by treatment arm.
Supplemental Table 4. Partially adjusted and fully adjusted intention-to-treat analysis results examining the effect of pantoprazole versus placebo on the outcomes of rate of eGFR decline—sensitivity analysis where those excluded from open-label extension because of kidney function (n=58) had their open-label extension eGFR imputed as 15 ml/min per 1.73 m2 and duration imputed as the mean of 3.3 years.
Supplemental Table 5. Subgroup analyses for the outcome of rate of eGFR change pantoprazole versus placebo using fully adjusted models (adjusted for age, sex, baseline eGFR, hypertension, diabetes, race, center, and rivaroxaban/ASA arm randomization).
Supplemental Table 6. Unadjusted intention-to-treat analysis results examining the effect of pantoprazole versus placebo on the outcomes of rate of eGFR change, CKD composite outcome, 40% eGFR decline, and AKI.
Supplemental Table 7. Partially adjusted and fully adjusted on-treatment analysis results examining the effect of pantoprazole versus placebo on the outcomes of rate of eGFR decline (ml/min per 1.73 m2 per year), CKD composite outcome, 40% eGFR decline, and AKI.
References
- 1.Rotermann M, Sanmartin C, Hennessy D, Arthur M. Prescription medication use by Canadians aged 6 to 79. Health Rep. 2014;25(6):3–9. PMID: 24941315 [PubMed] [Google Scholar]
- 2.Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. Trends in prescription drug use among adults in the United States from 1999-2012. JAMA. 2015;314(17):1818–1831. doi: 10.1001/jama.2015.13766 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Nguyen PV-Q, Tamaz R. Inappropriate prescription of proton pump inhibitors in a community setting. Can J Hosp Pharm. 2018;71(4):267–271. doi: 10.4212/cjhp.v71i4.2828 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Nardino RJ, Vender RJ, Herbert PN. Overuse of acid-suppressive therapy in hospitalized patients. Am J Gastroenterol. 2000;95(11):3118–3122. doi: 10.1111/j.1572-0241.2000.03259.x [DOI] [PubMed] [Google Scholar]
- 5.Othman F, Card TR, Crooks CJ. Proton pump inhibitor prescribing patterns in the UK: a primary care database study. Pharmacoepidemiol Drug Saf. 2016;25(9):1079–1087. doi: 10.1002/pds.4043 [DOI] [PubMed] [Google Scholar]
- 6.Zirk-Sadowski J Masoli JA Delgado J, et al. Proton-pump inhibitors and long-term risk of community-acquired pneumonia in older adults. J Am Geriatr Soc. 2018;66(7):1332–1338. doi: 10.1111/jgs.15385 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gomm W von Holt K Thomé F, et al. Association of proton pump inhibitors with risk of dementia: a pharmacoepidemiological claims data analysis. JAMA Neurol. 2016;73(4):410–416. doi: 10.1001/jamaneurol.2015.4791 [DOI] [PubMed] [Google Scholar]
- 8.Melloni C Washam JB Jones WS, et al. Conflicting results between randomized trials and observational studies on the impact of proton pump inhibitors on cardiovascular events when coadministered with dual antiplatelet therapy: systematic review. Circ Cardiovasc Qual Outcomes. 2015;8(1):47–55. doi: 10.1161/CIRCOUTCOMES.114.001177 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Perazella MA, Markowitz GS. Drug-induced acute interstitial nephritis. Nat Rev Nephrol. 2010;6(8):461–470. doi: 10.1038/nrneph.2010.71 [DOI] [PubMed] [Google Scholar]
- 10.Antoniou T Macdonald EM Hollands S, et al. Proton pump inhibitors and the risk of acute kidney injury in older patients: a population-based cohort study. CMAJ Open. 2015;3(2):E166–E171. doi: 10.9778/cmajo.20140074 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lazarus B Chen Y Wilson FP, et al. Proton pump inhibitor use and the risk of chronic kidney disease. JAMA Intern Med. 2016;176(2):238–246. doi: 10.1001/jamainternmed.2015.7193 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Xie Y, Bowe B, Li T, Xian H, Balasubramanian S, Al-Aly Z. Proton pump inhibitors and risk of incident CKD and progression to ESRD. J Am Soc Nephrol. 2016;27(10):3153–3163. doi: 10.1681/ASN.2015121377 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Klatte DCF Gasparini A Xu H, et al. Association between proton pump inhibitor use and risk of progression of chronic kidney disease. Gastroenterology. 2017;153(3):702–710. doi: 10.1053/j.gastro.2017.05.046 [DOI] [PubMed] [Google Scholar]
- 14.Tomlinson LA, Fogarty DG, Douglas I, Nitsch D. Pharmacoepidemiology for nephrologists: do proton pump inhibitors cause chronic kidney disease? Nephrol Dial Transplant. 2017;32(suppl 2):ii40–ii46. doi: 10.1093/ndt/gfw349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Baik SH, Fung K-W, McDonald CJ. The mortality risk of proton pump inhibitors in 1.9 million US seniors: an extended cox survival analysis. Clin Gastroenterol Hepatol. 2022;20(4):e671–e681. doi: 10.1016/j.cgh.2021.01.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Othman F, Crooks CJ, Card TR. Community acquired pneumonia incidence before and after proton pump inhibitor prescription: population based study. BMJ. 2016;355:i5813. doi: 10.1136/bmj.i5813 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Moayyedi P Eikelboom JW Bosch J, et al. Pantoprazole to prevent gastroduodenal events in patients receiving rivaroxaban and/or aspirin in a randomized, double-blind, placebo-controlled trial. Gastroenterology. 2019;157(2):403–412.e5. doi: 10.1053/j.gastro.2019.04.041 [DOI] [PubMed] [Google Scholar]
- 18.Moayyedi P Eikelboom JW Bosch J, et al. Safety of proton pump inhibitors based on a large, multi-year, randomized trial of patients receiving rivaroxaban or aspirin. Gastroenterology. 2019;157(3):682–691.e2. doi: 10.1053/j.gastro.2019.05.056 [DOI] [PubMed] [Google Scholar]
- 19.Bosch J Eikelboom JW Connolly SJ, et al. Rationale, design and baseline characteristics of participants in the cardiovascular outcomes for People using anticoagulation strategies (COMPASS) trial. Can J Cardiol. 2017;33(8):1027–1035. doi: 10.1016/j.cjca.2017.06.001 [DOI] [PubMed] [Google Scholar]
- 20.Eikelboom JW Connolly SJ Bosch J, et al. Rivaroxaban with or without aspirin in stable cardiovascular disease. New Engl J Med. 2017;377(14):1319–1330. doi: 10.1056/nejmoa1709118 [DOI] [PubMed] [Google Scholar]
- 21.US Food and Drug Administration. Research, C. For D. E. and. E6(R2) Good Clinical Practice: Integrated Addendum to ICH E6(R1); 2020. Accessed March 25, 2022. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/e6r2-good-clinical-practice-integrated-addendum-ich-e6r1 [Google Scholar]
- 22.Lewis JA. Statistical principles for clinical trials (ICH E9): an introductory note on an international guideline. Stat Med. 1999;18(15):1903–1942. doi: [DOI] [PubMed] [Google Scholar]
- 23.Kahan BC, Morris TP. Improper analysis of trials randomised using stratified blocks or minimisation. Stat Med. 2012;31(4):328–340. doi: 10.1002/sim.4431 [DOI] [PubMed] [Google Scholar]
- 24.Moayyedi P, Armstrong D, Hunt RH, Lei Y, Bukoski M, White RJ. The gain in quality-adjusted life months by switching to esomeprazole in those with continued reflux symptoms in primary care: EncomPASS--a cluster-randomized trial. Am J Gastroenterol. 2010;105(11):2341–2346. doi: 10.1038/ajg.2010.368 [DOI] [PubMed] [Google Scholar]
- 25.Jankowski JAZ de Caestecker J Love SB, et al. Esomeprazole and aspirin in Barrett’s oesophagus (AspECT): a randomised factorial trial. Lancet. 2018;392(10145):400–408. doi: 10.1016/S0140-6736(18)31388-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Moayyedi P. How to advise patients on the risk of chronic proton pump inhibitor therapy. Curr Opin Gastroenterol. 2020;36(4):317–322. doi: 10.1097/MOG.0000000000000641 [DOI] [PubMed] [Google Scholar]
- 27.Inker LA Heerspink HJL Tighiouart H, et al. GFR slope as a surrogate end point for kidney disease progression in clinical trials: a meta-analysis of treatment effects of randomized controlled trials. J Am Soc Nephrol. 2019;30(9):1735–1745. doi: 10.1681/ASN.2019010007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Oshima M Neal B Toyama T, et al. Different eGFR decline thresholds and renal effects of canagliflozin: data from the CANVAS program. J Am Soc Nephrol. 2020;31(10):2446–2456. doi: 10.1681/ASN.2019121312 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Xie Y, Bowe B, Li T, Xian H, Yan Y, Al-Aly Z. Long-term kidney outcomes among users of proton pump inhibitors without intervening acute kidney injury. Kidney Int. 2017;91(6):1482–1494. doi: 10.1016/j.kint.2016.12.021 [DOI] [PubMed] [Google Scholar]
- 30.Kellum JA.; Kidney Disease: Improving Global Outcomes (KDIGO) acute kidney injury work group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl 2012;2(1):1–138. doi: 10.1038/kisup.2012.1 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Bayer has set up a portal that allows users to submit proposals, which are then vetted by a committee and, if approved, data are made available for analysis.