Phosphodiesterase-5 inhibition and Alzheimer disease risk: A Mendelian randomization study

Background: Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the most common cause of dementia. Phosphodiesterase-5 (PDE-5) inhibitors such as sildenafil (Viagra), mainly used for erectile dysfunction as well as pulmonary arterial hypertension, have been highlighted as a potential therapeutic drug to prevent or treat AD. However, evidence of this causal relationship remains unclear.

Objectives: The aim of this study is to employ Mendelian randomization (MR) and colocalization methods to investigate the causal relationship between PDE-5 inhibition and the risk of AD.

Methods: We conducted a two-sample cis-MR analysis to study the link between genetically proxied PDE-5 inhibition and AD risk. Instruments, selected based on prior literature, fulfilled the following criteria: (1) were within the PDE5 gene; (2) were associated with PDE5 expression in blood (P < 5·10-8); (3) were pruned based on their diastolic blood pressure association (r2 < 0.35, window=10^4 kilobases); and (4) were tested for association with erectile dysfunction and pulmonary hypertension (positive control outcomes). We extracted instruments-exposure estimates from the blood pressure GWAS (N = 757,601 Europeans). Instrument-outcome estimates were obtained from an AD GWAS (N cases = 17,008, N controls = 37,154), which was selected based on SNP-heritability value and sample size. All alleles were harmonised with respect to the alleles of the linkage disequilibrium (LD) matrix (computed using the 1000 Genomes reference panel). MR estimates were calculated using the multiplicative random effects inverse-variance weighted method accounting for correlation between genetic instruments. Results were scaled to represent the diastolic blood pressure lowering effect of a 100mg dose of sildenafil. Colocalization was performed to determine if there were a shared causal variant between the two traits. Two sensitivity analyses were conducted: Two-Step cis-MR adjusting for traits associated with the instruments (detect it in Phenoscanner, P< 10-5); and 3 different GWAS summary statistics for AD, each with larger sample size than the one in the main analysis, but with extremely low SNP-heritability values.

Results: Five genetic variants were used as genetic instruments for PDE-5 inhibition. Results suggest no causal relation between genetically proxied inhibition of PDE-5 and AD (OR = 1.00, 95% confidence interval 0.96 to 1.04). Colocalization did not show evidence of a common causal variant (H4=0.9%). Two-step cis-MR results did not suggest genetic confounding in our results. MR estimates using other AD GWAS aligned with the main findings.

Conclusions: Our study suggests there is not a causal relationship between genetically proxied inhibition of PDE-5 and risk of developing AD.