Connecting proteomics and genomics to identify causal biomarkers specific for aortic valve stenosis

IntroductionAortic valve stenosis (AS) is a progressive disease characterized by the calcification and narrowing of the aortic valve, leading to significant morbidity and mortality. Early detection and risk stratification remain a major clinical challenge. Identifying specific biomarkers could significantly improve risk prediction and disease management. ObjectiveThis study aims to (1) identify novel plasma protein biomarkers for incident AS, (2) establish causal relationships using genetic approaches, and (3) prioritize biomarkers specific to the aortic valve tissue. MethodsWe assessed the association between 2,923 unique plasma proteins measured using the Olink Explore assay and AS incidence in 52,632 UK Biobank participants with 487 incident AS cases over a median follow-up of 13 years. Multivariable Cox proportional hazards models were used to evaluate associations adjusted for cardiovascular risk factors. A stratified analysis was performed to investigate the association in men and women separately. For causal inference, we used protein quantitative trait loci (pQTL) Mendelian Randomization (MR). We then verified the aortic valve specificity of the proteins using our transcriptomic dataset of 500 human aortic valves. Using expression QTL (eQTL)-based MR, we evaluated the causal role of gene expression in AS and performed colocalization analyses. ResultsOur fully adjusted model identified 55 proteins significantly associated with AS incidence, with GDF15 showing the strongest association (HR=1.93 per SD, 95% CI: 1.67-2.23, P value = 3.3E-19). In sex-stratified analyses, 10 proteins showed a significant association with AS incidence in women, including 5 more strongly associated in women (CD38, CD80, IGFBPL1, NFASC, SERPINA9, P interaction <0.05), whereas 16 proteins were identified in men including 1 protein (REG1A) with a significantly stronger association in men. pQTL-MR suggested a potential causal role for 4 proteins (PCSK9, CHI3L1, NFASC, PRSS8). Transcriptomic integration confirmed high aortic valve expression for 11 candidates, with three genes demonstrating significant associations in aortic valve eQTL MR analyses, including LTBP2, for which the pQTL and eQTL colocalized. ConclusionThe integration of proteomics and genomics allowed the identification of potential biomarkers and drug targets for AS, showing evidence of causality and tissue-specific expression.