Targeted Multiplex Proteomics For Prediction Of All-Cause Mortality During Long-Term Follow-Up In Outpatients With Peripheral Arterial Disease

Background and aims: Patients with peripheral arterial disease (PAD) are at high risk for fatal events. We aimed to investigate the ability among several serum proteins to predict all-cause mortality in outpatients with PAD.Methods: Consecutive outpatients with carotid and/or lower extremity PAD were included in the discovery cohort (n = 436), and subjects with PAD from a population-based sample in the validation cohort (n = 129). Blood samples were analyzed for 81 proteins by a proximity extension assay. The proteins best predicting incident all-cause mortality were identified using L1-regularized Cox regression. The added value of the identified proteins to clinical risk markers was evaluated by Cox regression models and presented by the area under the receiver operator characteristics curves (AUC).Results: In the discovery cohort (mean age 70 years; 59% men), 195 died (4.8 events per 100 person-years) during a 10.3 years median follow-up. The clinical risk markers generated an AUC of 0.70 (95% confidence interval [95%CI] 0.65-0.76). The two serum protein biomarkers with best prediction of all-cause mortality were growth differentiation factor 15 and tumor necrosis factor-related apoptosis-inducing ligand receptor 2. Adding these proteins to the clinical risk markers significantly improved prediction (p < 0.001) and yielded an AUC of 0.76 (95%CI 0.71-0.80). A higher discriminatory performance was observed in the validation cohort (AUC 0.84; 95% CI 0.76-0.92).Conclusions: In a large-sample targeted proteomics assay, we identified two proteins that improved risk prediction beyond the COPART risk score. The use of high-throughput proteomics assays may identify potential biomarkers for improved risk prediction in patients with PAD.