Plasma Proteome Clusters in Pulmonary Hypertension and Their Therapeutic Implications

<bold>Introduction:</bold> Pulmonary hypertension (PH) classification guides treatment, but some patients' conditions are complex. This study employs unsupervised plasma proteome clustering to develop a molecular PH classification. <bold>Methods:</bold> From 2013 to 2021, 470 PH patients, 136 disease controls, and 59 healthy controls from Imperial College Healthcare NHS Trust were recruited. Plasma levels of ~7000 proteins were assayed using the SomaScan-7K platform. Proteins differentiating PH from controls underwent unsupervised k-means clustering following UMAP (Uniform Manifold Approximation and Projection) dimensionality reduction. Clusters were compared clinically and validated in independent cohorts from the UK (n=165) and France (n=79). <bold>Results:</bold> Clustering of 156 distinct PH-associated proteins formed 4 clusters, including all clinical PH groups, with diverse 5-year survival rates: 78% (cluster 4), 62% (cluster 2), 44% (cluster 3), and 33% (cluster 1), p<0.001. Validation cohorts confirmed that clusters were associated with survival and risk scores, p<0.001. We investigated 2 experimental drug targets: the Platelet-Derived Growth Factor (PDGF) pathway was up-regulated in cluster 3 compared to other clusters and Transforming Growth Factor-β (TGF-β) pathway was up-regulated in cluster 1. <bold>Conclusion:</bold> Plasma proteomic profiles distinguish 4 PH clusters predicting prognosis with potential for guiding therapy, including PDGF and TGF-β strategies. <fig><object-id>erj;64/suppl_68/OA1874/F1</object-id><object-id>F1</object-id><object-id>F1</object-id><graphic></graphic></fig>