How the variability between computer-assisted analysis procedures evaluating immune markers can influence patients’ outcome prediction