Plasma proteomics profiles predict the risk of future aortic aneurysm and aortic dissection

Aortic aneurysms and aortic dissections (AA/AD) are serious vascular conditions that often progress without symptoms and are associated with high mortality, highlighting the need for improved tools to predict the occurrence. This study aims to identify plasma proteins that can predict the risk of future AA/AD events and to combine these biomarkers with traditional risk factors to construct risk prediction model.

We analyzed plasma proteomic data from 22 416 participants in the UK Biobank, measuring 2911 proteins using the Olink Explore proximity extension assay. Plasma proteomics data were analyzed using Cox regression and machine learning techniques. Proteins significantly associated with AA/AD risk were identified, and predictive models were constructed by integrating these biomarkers with traditional risk factors such as age, sex, and blood pressure.

The Cox regression models identified 25 proteins significantly associated with AA/AD risk, after adjusting for demographic factors. Furthermore, light gradient-boosting machine was used to rank the importance of these proteins and applied forward stepwise selection to identify four key predictive proteins (cystatin 3 [CST3], matrix metallopeptidase 12 [MMP12], multiple EGF-like domains 9 [MEGF9], and C-X-C motif chemokine ligand 17 [CXCL17]). The protein panel demonstrated an overall predictive AUC of 0.725 for AA/AD. The demographic model achieved an AUC of 0.740. Integration of these biomarkers with demographic factors significantly enhanced predictive accuracy, achieving an AUC of 0.777 (DeLong test P <0.001). Temporal trajectory analysis revealed that elevated levels of CST3, MMP12, and CXCL17 were detectable up to 10 years prior to AA/AD diagnosis.

Our study highlights the potential of plasma proteomics, particularly combination of four proteins (CST3, MMP12, MEGF9, and CXCL17), as a valuable strategy for predicting AA/AD risk. The integration of proteomic biomarkers with demographic factors enhances predictive accuracy and offers insights into the underlying molecular mechanisms, which could lead to improved early detection and personalized treatment for AA/AD.