Profiling EGFR isoforms in ovarian cancer demonstrates that EGFR and sEGFR have prognostic significance: Implications for EGFR isoforms as cancer biomarkers.

A58

Objectives: The epidermal growth factor receptor (EGFR) plays an important role in the etiology and progression of epithelial ovarian cancer (EOC). Previous immunohistochemical (IHC) studies have yielded discordant observations regarding EGFR as a prognostic factor of EOC, perhaps as a result of the inability to distinguish between full-length EGFR and one or more soluble EGFR (sEGFR) isoforms, which arise from alternate transcription. We, therefore, investigated whether sEGFR and EGFR are prognosticators of EOC.Methods: IHC was performed on a tissue microarray containing 196 EOC cases annotated with clinicopathological information using antibodies that discern EGFR from sEGFR. Clinicopathological covariates included: age, stage, grade, histology, bilateral ovarian involvement (BOI), synchronous or metachronous secondary cancers, and IHC labeling of p53, EGFR, and sEGFR. These covariates were analyzed as prognosticators of progression-free survival (PFS) and overall survival (OS) using accelerated failure time and Cox proportional hazards regression, respectively.Results: Univariate analyses confirm that age, stage, grade, histology, and BOI are significant prognosticators of PFS and OS in EOC patients. Multivariate regression shows that only older age, advanced stage, and BOI are independent, adverse prognosticators of both PFS and OS. In addition, high p53 labeling emerged as a favorable prognosticator of PFS (OR = 0.595; CI: 0.337, 1.049) and OS (HR = 0.637; CI: 0.431, 0.939) when mutually adjusted for confounding by age, stage, and BOI. Additional modeling revealed a significant interaction between EGFR and sEGFR for both PFS and OS. Adjusted for age, stage, BOI, and p53 labeling, patients with both high EGFR and sEGFR labeling had 1.898 (95% CI: 1.051, 3.428) higher odds of progression and 1.599 (95% CI: 1.203, 2.125) higher hazard of dying compared to patients with both low EGFR and sEGFR labeling.Conclusions: Adjusted for age, stage, BOI, and p53 and accounting for effect modification by EGFR and sEGFR, prominent labeling of both EGFR and sEGFR is a significant adverse prognosticator for EOC patients. These results have biological implications for the treatment of ovarian cancer patients, and suggest that prior studies examining the prognostic utility of EGFR using routine IHC methods may be confounded by the co-expression of alternate EGFR isoforms.