Background: T cell density in colorectal cancer (CRC) has proven to be of high prognostic importance. Here, we evaluated the influence of a hyperfractionated preoperative short-term radiation protocol (25 Gy) on immune cell density in tumor samples of rectal cancer (RC) patients and on patient survival. In addition, we assessed spatial tumor heterogeneity by comparison of analogue T cell quantification on full tissue sections with digital T cell quantification on a virtually established tissue microarray (TMA). Methods: A total of 75 RC patients (60 irradiated, 15 treatment-naïve) were defined for retrospective analysis. RC samples were processed for immunohistochemistry (CD3, CD8, PD-1, PD-L1). Analogue (score 0–3) as well as digital quantification (TMA: 2 cores vs. 6 cores, mean T cell count) of marker expression in 2 areas (central tumor, CT; invasive margin, IM) was performed. Survival was estimated on the basis of analogue as well as digital marker densities calculated from 2 cores (Immunoscore: CD3/CD8 ratio) and 6 cores per tumor area. Results: Irradiated RC samples showed a significant decrease in CD3 and CD8 positive T cells, independent of quantification mode. T cell densities of 6 virtual cores approximated to T cell densities of full tissue sections, independent of individual core density or location. Survival analysis based on full tissue section quantification demonstrated that CD3 and CD8 positive T cells as well as PD-1 positive tumor infiltrating leucocytes (TILs) in the CT and the IM had a significant impact on disease-free survival (DFS) as well as overall survival (OS). In addition, CD3 and CD8 positive T cells as well as PD-1 positive TILs in the IM proved as independent prognostic factors for DFS and OS; in the CT, PD-1 positive TILs predicted DFS and CD3 and CD8 positive T cells as well as PD-1 positive TILs predicted OS. Survival analysis based on virtual TMA showed no impact on DFS or OS. Conclusion: Spatial tumor heterogeneity might result in inadequate quantification of immune marker expression; however, if using a TMA, 6 cores per tumor area and patient sample represent comparable amounts of T cell densities to those quantified on full tissue sections. Consistently, the tissue area used for immune marker quantification represents a crucial factor for the evaluation of prognostic and predictive biomarker potential.
Automatic Clump Splitting for Cell Quantification in Microscopical Images