Metastasizing breast cancer is an incurable disease. Thus, it is important to find proteins responsible for tumor dissemination, which may be used as biomarkers or treatment targets. In this study, we established the combination of a non-targeted LC-MS/MS discovery and a targeted LC-SRM verification workflow for improved discovery and validation of biomarkers. We collected 80 breast tumors, stratified for estrogen receptor status and development of distant recurrence (DR+/-). After enrichment of N-glycosylated proteins, label-free LC-MS/MS was performed on all of the individual tumors in triplicates. In total, 1515 glycopeptides from 778 proteins were identified, used to create a map of the breast cancer N-glycosylated proteome. Based on the breast cancer N-glycosylated proteome map we constructed a 92-plex targeted label-free LC-SRM panel, enriched for interesting proteins, and analyzed the samples for the selected proteins, resulting in 10 proteins differentially regulated between DR+/DR- tumors. We also compared the LC-SRM results to clinically reported HER2 status. 17 of 18 patients could be classified the same way, demonstrating the clinical accuracy of LC-SRM. In conclusion, we demonstrated the use of a combined non-targeted LC-MS/MS and targeted LC-SRM strategy, at large scale on clinical samples, and identified 10 proteins as potential biomarkers.