ABSTRACT
BACKGROUND: Better biomarkers must be found to develop clinically useful urine tests for bladder cancer. Proteomics can be used to identify the proteins released by cancer cell lines and generate candidate markers for developing such tests. METHODS: We used shotgun proteomics to identify proteins released into culture media by eight bladder cancer cell lines. These data were compared with protein expression data from the Human Protein Atlas. Epidermal growth factor receptor (EGFR) was identified as a candidate biomarker and measured by ELISA in urine from 60 noncancer control subjects and from 436 patients with bladder cancer and long-term clinical follow-up. RESULTS: Bladder cancer cell lines shed soluble EGFR ectodomain. Soluble EGFR is also detectable in urine and is highly elevated in some patients with high-grade bladder cancer. Urinary EGFR is an independent indicator of poor bladder cancer-specific survival with a hazard ratio of 2.89 (95% CI 1.81-4.62, P<0.001). In multivariable models including both urinary EGFR and EpCAM, both biomarkers are predictive of bladder cancer-specific survival and have prognostic value over and above that provided by standard clinical observations. CONCLUSIONS: Measuring urinary EGFR and EpCAM may represent a simple and useful approach for fast-tracking the investigation and treatment of patients with the most aggressive bladder cancers.