675P - A clinico-molecular risk stratification model for resected gastric cancer: prognostic impact of HER2, FHIT and APC expression status

Date 30 September 2012
Event ESMO Congress 2012
Session Poster presentation II
Topics Biomarkers
Gastric Cancer
Presenter Emilio Bria
Authors E. Bria1, G. De Manzoni2, S. Beghelli3, A. Tomezzoli4, S. Barbi4, M. Frizziero1, I. Sperduti5, S. Bersani4, G. Tortora6, A. Scarpa7
  • 1Medical Oncology, Azienda Ospedaliera Universitaria Integrata Verona-"Borgo Roma", 37134 - Verona/IT
  • 21st Division Of General Surgery, University of Verona, Verona/IT
  • 3Pathology, ARC-NET Applied Research on Cancer Center, Verona/IT
  • 4Pathology And Diagnostics, University of Verona, Verona/IT
  • 5Biostatistics, Regina Elena Institute, Roma/IT
  • 6Oncologia Medica, Azienda Ospedaliera Universitaria Integrata Verona-"Borgo Roma", 37134 - Verona/IT
  • 7Patologia E Diagnostica, Anatomia Patologica, University of Verona, Verona/IT



In order to complement the prognostic power of clinical parameters for resected gastric cancer, a stratification model to predict individual patient risk was developed taking into account 11 molecular factors.


Clinicopathological and molecular data (expression of Cdx2, Apc, ß-Catenin, E-Cadherin, Fhit, p53, Her2; HER2 and TOPO2A gene copy number; PIK3CA mutations; microsatellite instability-MSI) were correlated to cancer-specific/overall survival (CSS/OS) using a Cox model. A logistic equation including regression analysis coefficients was constructed to estimate individual patients' probability (IPP) of death. Internal cross-validation (100 simulations, 80% of the dataset) was accomplished. Ratios from the multivariate model served to derive a prognostic continuous score to identify risk classes.


Two-hundred-eight patients were studied (median follow-up 20 months). At multivariate analysis, sex (HR 1.53, p = 0.04), stage (HR 5.40, p < 0.0001), margins (HR 2.69, p < 0.0001), localization (HR 1.64, p = 0.008), resected nodes (HR 1.55, p = 0.02), APC (HR 1.91, p = 0.001), FHIT (HR 1.54, p = 0.05) and HER-2 (HR 1.92, p = 0.08), were independent predictors for CSS; the same factors (plus age and except Fhit) predicted OS. Multivariate model predicted IPP with high prognostic accuracy (0.87 for CSS; 0.91 for OS). A 2-class risk stratification model, developed on the basis of the ROC generated cut-off prognostic score (AUC 0.87; Sensitivity 85.3%, Specificity 78.9%), significantly separated low and high risk patients for CSS (16.4% and 82.5%, p < 0.0001) and OS (15.6% and 79.7%, p < 0.0001), with a prognostic performance of 0.82 for CSS and 0.81 for OS. A further 3-class-model differentiated low, intermediate, and high-risk patients for CSS (7.3%, 37.3%, and 89.7%, p < 0.0001) and OS (0%, 23.3%, and 84.8%, p < 0.0001).


The concurrent evaluation of the expression of Apc, Fhit, Her2 proteins and clinicalpathological parameters (Stage, Resected Nodes, Margins, Location, Sex) may powerfully discriminate the prognosis of resected gastric cancer patients.


All authors have declared no conflicts of interest.