Proton Pump Inhibitors, Nephropathy, and Cardiovascular Disease in Type 2 Diabetes | 03:23 |
Proton Pump Inhibitors, Nephropathy, and Cardiovascular Disease in Type 2 Diabetes The Fremantle Diabetes Study Timothy M. E. Davis; Jocelyn Drinkwater; Wendy A. Davis DISCLOSURES J Clin Endocrinol Metab. 2017;102(8):2985-2993 Abstract and Introduction . Abstract Context: There is emerging evidence of various adverse effects of chronic proton pump inhibitor (PPI) therapy. Objective: To assess the impact of PPI use on nephropathy and cardiovascular disease (CVD) risk in type 2 diabetes. Design: Longitudinal observational study. Setting: Urban-dwelling community. Patients: Patients with type 2 diabetes from the Fremantle Diabetes Study Phase II and on stable renin-angiotensin system blocking therapy were divided into those remaining untreated with a PPI (group 1, n = 686), on PPI therapy throughout (group 2, n = 174), and commencing (group 3, n = 109) or discontinuing regular PPI therapy (group 4, n = 67) during the 2 years between assessments. Main Outcome Measures: Changes (Δ) in urinary albumin/creatinine ratio (uACR), estimated glomerular filtration rate (eGFR), and predicted 5-year CVD risk. Results: There were no statistically significant differences in ΔuACR between groups [analysis of variance (ANOVA), P = 0.36], but ΔeGFR was different (ANOVA, P = 0.002), with group 3 exhibiting a greater reduction than group 1 [adjusted mean difference (95% confidence interval), −2.7 (−4.5 to −0.8) mL/min/1.73 m2; P = 0.005]. The Δ5-year CVD risk showed a similar pattern (ANOVA, P < 0.001), with group 3 having a greater increase than group 1 [adjusted mean difference (95% confidence interval), 1.7% (0.6% to 2.8%); P = 0.002]. Conclusions: Although PPI use was not associated with a sustained adverse effect on uACR, the association between PPI initiation and both worsening nephropathy and increasing 5-year CVD risk has potential clinical implications in type 2 diabetes. Introduction A recent in vitro study found that prolonged exposure to the proton pump inhibitor (PPI) esomeprazole impaired endothelial function and accelerated human endothelial senescence by reducing telomere length.[1] The authors postulated that irreversible endothelial damage associated with long-term PPI use could be a unifying mechanism underlying the observation that the PPI class of drugs is associated with cardiovascular disease (CVD), cognitive decline, and renal impairment in administrative database and pharmacovigilance studies.[2–5] Diabetes is a disease characterized by endothelial dysfunction and premature vascular aging,[6] and the adverse endothelial effects of PPIs[1] could accelerate diabetes-associated angiopathy and contribute to complications and death. Albuminuria is considered to reflect widespread endothelial damage[7,8] and could be used as an in vivo marker of PPI effects on the vasculature.[9] It can be hypothesized, therefore, that patients with type 2 diabetes initiating PPI therapy will manifest a subsequent increase in urinary albumin excretion that will persist after withdrawal of treatment. We have examined this hypothesis and also assessed PPI effects on renal function and cardiovascular risk by using longitudinal observational data from the representative community-based Fremantle Diabetes Study Phase II (FDS2). Materials and Methods Patients and Approvals The FDS2 is a prospective cohort study of 1732 participants recruited between 2008 and 2011 from a postcode-defined urban population of 157,000 in the Australian state of Western Australia.[10] Of these, 1551 (89.5%) had type 2 diabetes [mean±standard deviation (SD) age 65.7±11.6 years, 51.9% males]. Details of recruitment, characterization of diabetes type, and nonrecruited patients (who were similar to recruits in age, sex distribution, and diabetes type) have been published.[10] The FDS2 protocol was approved by the South Metropolitan Area Health Service Human Research Ethics Committee. Written informed consent was obtained from each participant. Clinical Procedures Each patient underwent a detailed assessment at study entry that comprised a comprehensive questionnaire, physical examination, and fasting biochemical and hematologic tests conducted in a single nationally accredited laboratory (PathWest Laboratory Medicine, Fremantle Hospital). Similar biennial face-to-face assessments interspersed with biennial postal questionnaires are continuing with all patients having been scheduled for at least two biennial (years 2 and 4) reviews. Details of all medications are recorded at each assessment. Chronic complications are ascertained using standard criteria, and CVD events are ascertained by self-report and validated linkage with government morbidity and mortality databases.[10] Albumin and creatinine concentrations were measured in serum and first-morning urine samples using an Architect ci8200 (Abbott Laboratories, Abbott Park, IL) with betweenday coefficients of variation of <5.0%.[11] The estimated glomerular filtration rate (eGFR) was determined using the Chronic Kidney Disease Epidemiology Collaboration equation.[12] The urinary albumin/creatinine ratio (uACR) was calculated from the respective urinary concentrations. Data Analysis We analyzed data from FDS2 patients who 1) remained off PPI therapy throughout follow-up (group 1), 2) remained on PPI therapy throughout follow-up (group 2), 3) commenced regular therapy with a PPI and who had a biennial assessment either side of this therapeutic change (group 3) with an average of duration of therapy of 12 months if the time of initiation was random during the follow-up period, or 4) were taking a PPI and had discontinued the medication at the next face-to-face review (group 4) with an average of duration off therapy of 12 months if the time of cessation was random during follow-up (see Figure 1). In the case of groups 1 and 2, comparisons were between baseline and year 2 data. Group 3 comparisons were made between either baseline and year 2 or between year 2 and year 4 depending on when the PPI was initiated. Similarly, group 4 comparisons were made between either baseline and year 2 or between year 2 and year 4 depending on when the PPI was stopped. Groups 3 and 4 contained a small number of patients who restarted after stopping or who stopped after starting (see Figure 1). We excluded[1] patients who did not have valid biochemical testing at both assessments and, because angiotensin-converting enzyme inhibitor and angiotensin receptor blocker therapies can influence both uACR and eGFR,[13,14] and[2] those whose angiotensin-converting enzyme inhibitor/angiotensin receptor blocker treatment status changed between time-points (i.e., who started or stopped these medications). Click to zoom (Enlarge Image) Figure 1. Consort diagram showing the numbers of the patients recruited to the FDS2 and those selected for the present subgroup analyses. ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; T2DM, type 2 diabetes mellitus. The computer package IBM SPSS Statistics 22 (IBM Corp., Armonk, NY) was used for statistical analysis. Data are presented as proportions, mean ± SD, geometric mean (SD range), or, in the case of variables that did not conform to a normal or lognormal distribution, median and interquartile range. For independent samples, two-way comparisons for proportions were by Fisher's exact test, for normally distributed variables by Student t test, and for nonnormally distributed variables by Mann-Whitney U test. Multiple logistic regression with forward conditional variable selection (P < 0.05 for entry, P > 0.10 for removal) of clinically plausible variables with bivariate P<0.20 was used to identify independent associates of PPI use. A validated prediction equation based on outcome data from the Fremantle Diabetes Study Phase I[15] was used to estimate the 5-year risk of a major adverse CVD event (myocardial infarction, stroke, or CVD death) using the baseline variables age, sex, race/ethnicity (Aboriginal or Southern European), hemoglobin A1c, serum high-density lipoprotein cholesterol, uACR, and a history of CVD. Changes in normally distributed variables and in logtransformed (ln) variables, including uACR and CVD risk, were assessed using the paired t test. Mean changes in untransformed variables over time were compared by PPI group using analysis of variance (ANOVA) and multiple linear regression with group 1 as reference. Initial parsimonious models were obtained by comparing clinically plausible baseline confounding variables by PPI group and considering those with bivariate P < 0.20 for entry into forward stepwise multiple linear regression models (P < 0.05 for entry, P > 0.10 for removal). Mean changes were then compared by PPI group with group 1 as reference after adjusting for the respective most parsimonious models. In addition, between-group differences in uACR were assessed after adjustment for all baseline variables that differed across the PPI groups at P<0.20 and not just those in each individual most parsimonious model. Results The characteristics of FDS2 participants with type 2 diabetes are summarized in Table 1, categorized by PPI treatment status at study entry. There were 342 taking a PPI at baseline (esomeprazole, 36.7%; pantoprazole, 24.9%; omeprazole, 21.1%; rabeprazole, 12.6%; and others, 4.7%). Those taking a PPI were older at diagnosis and study entry, and they had a longer duration of diabetes than the 1209 who were not. They were more likely to be obese and treated for hypertension but had better glycemic control. They had better serum lipid profiles in the presence of more frequent lipid-lowering therapy. There was no statistically significant betweengroup difference in uACR or albuminuria category, but the patients on a PPI were more likely to have at least stage 2 renal impairment. They were also more likely to have evidence of CVD and to be treated with a nonsteroidal anti-inflammatory drug. These bivariate differences were consistent with the independent associates of PPI use in logistic regression analysis. Variables inversely associated with PPI use were an eGFR ≥90 mL/min/1.73 m2 [odds ratio (OR) and 95% confidence interval, 0.60 (0.45 to 0.80); P < 0.001], hemoglobin A1c [0.84 (0.76 to 0.94) for an increase of 1% or 11mmol/mol; P = 0.002], male sex [0.71 (0.55 to 0.92); P = 0.009], education beyond primary level [0.66 (0.47 to 0.94); P = 0.019], and an Aboriginal racial background [0.40 (0.18 to 0.91); P = 0.028]. Variables positively associated with PPI use were cerebrovascular disease [2.12 (1.42 to 3.16); P < 0.001], coronary heart disease [1.62 (1.23 to 2.13); P = 0.001], nonsteroidal anti-inflammatory drug therapy [1.60 (1.11 to 2.30); P = 0.011], and lipidmodifying treatment [1.42 (1.06 to 1.92); P = 0.021]. There were 1016 participants eligible for the longitudinal study (20 were counted twice due to commencement of a PPI at year 2 and then discontinuation at year 4 or vice versa). Their characteristics at FDS2 entry are summarized by PPI treatment status in Supplemental Table 1. They were not statistically significantly different from the 535 ineligible participants by age at FDS2 entry, sex, or PPI use (P > 0.06) but had shorter diabetes duration [median (interquartile range) 8.0 (2.0 to 15.3) vs 10.0 (3.0 to 17.0) years; P = 0.001] and were less likely to have died during follow-up to the end of year 4 (8.2%vs 20.4%; P<0.001). The baseline characteristic of the four groups of eligible participants are summarized in Table 2. Those in group 1 tended to be younger and to have a shorter diabetes duration than participants in the other three groups. They also tended to be less intensively treated with antihypertensive and lipid-modifying therapies and to have a higher eGFR, consistent with a less frequent history of CVD and a lower 5-year CVD risk score. The overall mean ± SD time between the two FDS2 assessments was 2.1±0.3 years. The variables of interest at baseline and follow-up are summarized in Table 3, together with the temporal changes after adjustment in parsimonious models in which variables in Table 2 that differed across the four groups with P < 0.20 were considered for entry. There was no temporal change in geometric mean uACR in groups 1 and 2 (P ≥ 0.47). There was a trend to an increase in group 3 (P = 0.056) and a statistically significant decrease in group 4 (P = 0.015). Compared with group 1 (reference), the adjusted mean changes from baseline (ΔuACR) in each of groups 2, 3, and 4 were not statistically significantly different (P ≥ 0.15; see Supplemental Figure 1). The between-group differences in uACR were similar after adjustment for all baseline variables that differed across the PPI groups except that the increase in group 4 vs group 1 patients was statistically significant (see Supplemental Table 3). In a parallel analysis of eGFR, there was a statistically significant temporal reduction in eGFR in groups 1, 2, and 3 (P < 0.001 in each case) but no change in those who stopped PPI therapy (group 4; P = 0.45). In the case of ΔeGFR, there were statistically significant differences between the groups (P = 0.002 by ANOVA), with those who initiated a PPI (group 3) exhibiting a statistically significantly greater adjusted mean reduction compared with group 1 (−2.7 mL/min/1.73 m2; P = 0.005; see Supplemental Figure 1). This patternwas also seen in a model that adjusted for all baseline variables that differed across all PPI groups at P < 0.20 (see Supplemental Table 3). There was a statistically significant increase in composite 5-year CVD risk score in each of the four groups during the mean 2.1 years of follow-up (P < 0.001; see Table 3 and Supplemental Table 2 for group-specific changes in individual CVD risk factors). The Δ5-year CVD risk score was statistically significantly different between the four groups (P < 0.001 by ANOVA), with those who initiated a PPI (group 3) exhibiting a greater adjusted mean increase compared with group 1 (1.7%; P = 0.002; see Supplemental Figure 1). These findings parallel preliminary analysis of largely self-reported CVD events (coronary heart disease, cerebrovascular disease, or CVD death) between the two FDS2 biennial assessments in the four groups of patients (2.8%, 5.2%, 9.2%, and 4.5% in groups 1 through 4, respectively; P = 0.012). After adjustment for age, sex, and diabetes duration, and using group 1 as reference, only group 3 had a statistically significant OR (95% confidence interval) for CVD[3.67 (1.63 to 8.23), P = 0.002 vs 2.02 (0.88–4.60) and 1.71 (0.49–5.97) for groups 2 and 4, respectively;P>0.10 in each case]. The adjusted between-group differences in 5-year CVD risk score were similar after adjustment for all baseline variables that differed across all PPI groups at P<0.020 (see Supplemental Table 3). Discussion The present real-life longitudinal data suggest that initiation of a PPI in community-based patients with type 2 diabetes during an average 2.1-year period of follow-up does not result in a clinically significant and sustained increase in uACR relative to that in other patients. However, PPI initiation may accelerate the diabetesassociated decline in renal function and increase 5-year CVD risk over the same time. Data from our patients who stopped PPI therapy during follow-up provide evidence that the deleterious effects associated with this group of drugs do not persist. These data, which are, to our knowledge, the first that address the potential adverse effects of PPI therapy in diabetes, may have implications for the management of type 2 diabetes outside glycemia and nonglycemic CVD risk factors. The patients taking a PPI at FDS2 entry had better glycemic control (lower fasting serum glucose and hemoglobin A1c) than those who were not despite similar blood glucose-lowering therapies. This finding is in accord with a range of other studies showing that PPIs improve glycemia in diabetes, probably through the trophic effects of increased circulating gastrin concentrations on the pancreatic β-cell mass.[16] Consistent with general population studies,[17,18] baseline PPI use in the current study was more frequent in those with limited education and with greater cardiovascular comorbidity and its treatment, including lipid-lowering therapy. This could also reflect use of PPIs to reduce the risk of gastrointestinal hemorrhage associated with use of aspirin and other nonsteroidal anti-inflammatory drugs.[19] The association between abdominal obesity and increased risk of gastroesophageal reflux disease and thus PPI use is well recognized.[20] We postulate that the inverse association between Aboriginal racial background and PPI use indicates access issues consistent with continued socioeconomic disadvantage.[11] We interpret the lack of a statistically significant influence of initiation of a PPI and of chronic PPI use on uACR as suggesting that the permanent and marked deleterious effects of this group of drugs on endothelial function found in in vitro studies[1] are not replicated in vivo. This is consistent with the results of a short-term study of 21 adults without diabetes, approximately half of whom had CVD.[21] Nonsignificant trends in the present uACR data suggested that those initiating a PPI (group 3) had the greatest temporal increase in uACR but that those on chronic PPI therapy (group 2) had similar ΔuACR as those who remained untreated (group 1), whereas the mean ΔuACR in those stopping the medication (group 4) was negative (a finding that became significant after adjustment for all baseline variables that differed across the PPI groups). Although these findings parallel the more pronounced temporal changes in eGFR, overall they suggest that there is no clinically significant chronic progressive change in endothelial function when PPIs are prescribed in patients with type 2 diabetes. There was a strong independent inverse association between PPI use and renal function at study entry in our patients. This is likely to represent a nephropathic effect of PPIs that has been found in a number of large-scale administrative database and pharmacovigilance studies[3,22–24] and is also consistent with the changes across the present four groups during follow-up. Group 3 patients had the greatest decline in eGFRover the mean 2.1 years of follow-up, a mean fall that was approaching three times that in untreated group 1 patients. The fact that group 2 patients had a similar ΔeGFR to that in group 1 might suggest that chronic use is associated with an adaptive response, and it is of relevance that there was evidence of an attenuation of the adverse effect of PPI therapy of more than 2 years' duration in a recent meta-analysis.[22] It could also represent patient selection with those experiencing adverse (including renal) effects on PPIs having withdrawn from treatment before FDS2 entry. Nevertheless, group 4 patients had a slower rate of eGFR decline than would be expected for patients with type 2 diabetes, suggesting potential reversibility on cessation of therapy. The PPI group of drugs is known to cause acute interstitial nephritis, which can progress subclinically to chronic interstitial nephritis.[3] Most cases of acute interstitial nephritis are reversible, but the development of fibrosis may herald an irreversible progression to endstage renal disease,[25] a situation that emphasizes the importance of an awareness of the nephrotoxic effects of PPIs, especially in a vulnerable population such as patients with type 2 diabetes. One animal study has, by contrast, shown that lansoprazole is associated with renoprotection,[26] an effect attributed to the antiinflammatory and antioxidant properties of PPIs,[27] but the relevance to human disease is uncertain given the present and other data. Because uACR is strongly associated with incident CVD[7–9] and a variable in our validated FDS CVD risk equation,[15] and because PPIs may have effects on glycemia,[16] we assessed 5-year CVD risk in the same way as uACR and eGFR. Patients in group 3 had the greatest increase in the predicted risk of an incident CVD event, a result that tallied with the between-group differences in ORs for largely self-reported incidence of CVDevents during the approximate 2 years of follow-up. As with eGFR, we cannot rule out selection or survivor bias resulting in temporal changes in CVD risk in chronically treated group 2 patients that were no different from those on no PPI therapy or withdrawing from a PPI during follow-up. The significant increase in 5-year CVD risk in group 3 patients persisted after adjustment for potentially confounding variables and accords with the association between PPIs and CVD end points found in other studies.[4,5] Evidence of mitigating pleiotropic anti-inflammatory and antioxidant properties[27] has not been confirmed in all studies,[28] whereas PPI effects on the absorption of micronutrients such as magnesium[29] may have adverse consequences for CVD independent of direct drug effects.[30] The FDS2 is a prospective study with ongoing collection of morbidity and mortality data that should allow a future assessment of the relationship between PPI use and CVD events and death in type 2 diabetes. The current study had limitations. We examined associations between PPI use over at least 2 years and key outcomes, but the data cannot be used to infer causality. Although our findings do not suggest, as preempted by preclinical evidence,[1] that there is a large permanent effect of PPI on endothelial function in patients with type 2 diabetes, a small but clinically significant change may have been detected with a larger patient sample followed for a longer period, especially those in group 4 who stopped PPI therapy during follow-up. We did not have measurements of key variables between FDS2 assessments, including at the time of initiation or cessation of PPI therapy in groups 3 and 4, respectively. As acknowledged, selection and survival bias may have influenced the results, especially for those in group 2 on chronic PPI therapy. We had relatively small numbers of CVD outcomes, but accumulation of more robust and complete data is in progress through validated linkage. The strength of the current study is its access to near-complete prospective data froma large wellcharacterized community-based sample. The PPI group of drugs has radically changed the management of acid-related diseases in recent decades, but these drugs are expensive and associated with adverse effects when used for prolonged periods.[31] The present data provide a note of caution in the context of type 2 diabetes. Regular assessment of their appropriateness as part of chronic pharmacotherapy, as well as regular monitoring of renal function during chronic use, appears important, and there is an argument that CVD risk factor management should be optimized if PPIs are prescribed given the present association with CVD risk after their initiation. Acknowledgments We are grateful to the FDS2 participants for their involvement and FDS2 staff for help with collecting and recording clinical information. We thank the Biochemistry Departments at Fremantle Hospital and Health Service and PathWest Laboratory Medicine, Sir Charles Gairdner Hospital, for performing laboratory tests. Abbreviations ANOVA, analysis of variance; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; FDS2, Fremantle Diabetes Study Phase II; OR, odds ratio; PPI, proton pump inhibitor; SD, standard deviation; uACR, urinary albumin/creatinine ratio. J Clin Endocrinol Metab. 2017;102(8):2985-2993. | |
Просмотров: 1963 |
Добавил: galinadoctor
| Теги: |
Всего комментариев: 0 | |