Sex-specific prediction models for aortic aneurysm integrating traditional clinical risk factors and proteomic profiles: a large-scale prospective study from the UK Biobank

BACKGROUND: This study aimed to systematically map the sex-specific clinical and proteomic risk profiles of aortic aneurysm (AA), elucidate its molecular mechanisms, and develop a sex-specific protein risk prediction score. METHODS: Based on the UK Biobank, we adopted a sex-stratified strategy to assess associations between traditional clinical factors and AA in 471,660 participants, and performed proteomics analysis on 49,887 participants with plasma protein data. Mediation analysis was used to explore the molecular mechanisms by which clinical risks drive AA. Finally, sex-specific protein risk scores were developed via LASSO regression, and their predictive performance was evaluated in an independent validation cohort. RESULTS: Smoking (HR: Male: 2.82; Female: 4.42) and valvular disease (HR: Male: 2.01; Female: 4.62) were the strongest shared risk factors, with women exhibiting significantly higher susceptibility. Incident AA was primarily attributed to smoking (Male: 15.3%; Female: 19.8%) and hypertension (Male: 11.2%; Female: 10.2%). Smoking was associated with AA potentially through the ECM degradation pathway in both sexes. Hypertension may also influence AA risk through this pathway in men, whereas in women, it may primarily operate through metabolic and growth factor regulation pathways. Among LASSO-selected proteins, 4 were shared, 10 were male-specific, and 9 were female-specific. Ultimately, the model integrating traditional risk factors and sex-specific protein scores demonstrated superior predictive performance in an independent cohort (C-statistic: Male: 0.809; Female: 0.832). CONCLUSIONS: Smoking and hypertension are primary risk factors for AA. Men may be predisposed to structural destruction potentially mediated by ECM degradation, whereas women may be predisposed to intrinsic failure potentially involving metabolic dysregulation and cell apoptosis. The model integrating age, clinical factors, and protein scores better captures residual risk, significantly improving AA prediction.