Abstract
Vascular endothelial growth factor A (VEGF-A) is a very important growth factor in angiogenesis and holds potential as both a predictive marker for anti-angiogenic cancer treatment and a prognostic variable. Consequently, reliable estimation of VEGF expression is crucial. Investigators immunostained whole tumor sections for VEGF-A, VEGF-B, and VEGFR-1 of invasive ductal carcinomas of the breast and scored the tumors manually with staining intensity as the only parameter and by a combination of qualitative and quantitative information. The investigators also introduce an automated method for analyzing VEGF expression (so-called AI score) using the same tumor sections. Analysis of 100% of the tumor area was performed and the results were compared with the reduced analysis of 25% of the tumor area. These analyses were performed at ×5 and ×10 magnification, and each analysis was repeated in a second run with a new delineation of the tumor area. The AI scores were correlated to the manual scoring of VEGF intensity, but reproducibility of manual IHC scores was rather poor. The AI scores were reproducible, and the restricted analysis of 25% of the tumor area at ×5 magnifications was the most efficient considering time consumption and data load.