This study aimed to identify plasma proteins causally associated with pulmonary arterial hypertension (PAH) by using a proteome-wide Mendelian randomization (PWMR) approach. We integrated two large proteomic genome-wide association study (GWAS) data sets with two PAH GWAS data sets and applied PWMR, Bayesian colocalization, and SMR to prioritize candidate proteins. Drug prediction, molecular docking, single-cell RNA sequencing (scRNA-seq) profiling, and a phenome-wide association study (PheWAS) were further used to assess druggability and potential phenotypic associations. PWMR identified 271 PAH-associated proteins, which were refined to the following three targets: CASP10, NOTCH3, and NCAM2. Molecular docking showed strong predicted binding between CASP10 and usnic acid and masoprocol and between NOTCH3 and 2-mercaptobenzothiazole. scRNA-seq data revealed CASP10 expression in endothelial and inflammatory cells and NOTCH3 expression in smooth muscle cells and fibroblasts. The PheWAS did not detect any adverse associations that survived the false discovery rate (FDR) correction. Overall, this integrated analysis highlights CASP10 and NOTCH3 as potential therapeutic targets for PAH. CASP10, as supported by stronger predicted druggability and multiomics evidence, warrants priority for further experimental evaluation.