The effect of proton pump inhibitors on cognition and brain volume in the Rhineland study

Abstract

Introduction Proton pump inhibitors (PPI) are one of the most widely used drug classes in the world. Several studies have found that users of PPIs had an increased risk of cognitive decline and dementia. This could however not be confirmed by others. Protopathic bias, in which early signs and symptoms of the disease influence drug exposure, could be a reason for these conflicting results. Given that alterations in brain volume can be seen years before the occurrence of cognitive symptoms, examining the association between PPI use and brain volume in non-demented persons would eliminate the potential of protopathic bias. Therefore, we aimed to assess the associations between PPI use and cognition and brain volume. Methods We performed a cross-sectional study based on the first 1306 participants in the Rhineland Study. The Rhineland Study is a population-based cohort study in Bonn Germany that includes people aged ≥ 30 years. Information on medication use is collected in an interview in which regularly and as needed used medication is registered. PPI use was classified as regular use of PPIs (ATC code A02BC). Participants who reported using PPIs as needed were excluded. We assessed the following cognitive domains: processing speed [Trail Making Test (TMT) A], executive functioning (TMT B - A), working memory (Digit Span Forward), and long-term memory [Verbal Learning and Memory Test (VLMT); delayed loss]. Measures for TMT and VLMT were inverted so that higher values represent better test performance for all tests. All eligible participants were scanned on Siemens 3 Tesla Prisma MRI scanners. We assessed the effect of PPI use on total brain volume, cortical grey matter volume, and ventricle volume. Processing speed and ventricle volume were log transformed. All outcome measures were z-standardized. We examined the association between PPI use and cognition using multivariable linear regression models. Within the first model we only controlled for age (m1: age). Because PPI use strongly increased with increasing age, we included variables that were a determinant for PPI use after controlling for age in the fully adjusted model (m2: age, body mass index, diabetes, hypertension, antithrombotic use, statin use and NSAID use). We subsequently examined the association between PPI use and brain volume outcomes in the same way but with additional control for head size in all models. Results We included 1130 participants [mean age 55, standard deviation (SD) 13.5; 57% women; 81 PPI users] to determine the effect of PPI use on cognition. We excluded 176 participants, 25 for missing drug information, 89 for as needed PPI use and 62 for missing cognition data. We observed no effect on processing speed with a mean SD difference [95% confidence interval (95% CI)] for m1 of 0.00 (−0.19; 0.19) and for m2 of 0.02 (−0.18; 0.22). We observed a trend towards a negative effect of PPI use with a mean SD difference (95% CI) for executive functioning, working memory and long-term memory for m1 of −0.18 (−0.40; 0.03), −0.17 (−0.40; 0.05), and −0.29 (−0.51; −0.07) and for m2 of −0.13 (−0.35; 0.10), −0.12 (−0.36; 0.11), and −0.17 (−0.39; 0.06). Effect sizes were comparable to a 5-year increase in age for executive functioning and a 9-year increase in age for working memory and long-term memory. We had MRI data in a subset of 640 participants (mean age 54, SD 13.2; 56% women; 28 PPI users). Brain volume measures were smaller in PPI users compared to non-users, but these effects were far from significant, mean SD difference (95% CI) for total brain volume, cortical grey matter volume, and ventricle volume for m1 was −0.07 (−0.23; 0.09), −0.07 (−0.26; 0.12), and 0.02 (−0.27; 0.31), and for m2 was −0.05 (−0.21; 0.12), −0.03 (−0.23; 0.16) and −0.06 (−0.36; 0.24). Conclusions Given our small sample size and relatively young population, these results suggest a possible effect of PPI use on brain structure and function and warrant further investigation in larger samples.