In recent years, great progress has been made in the development of targeted therapy against cancer. However, it is also more recognized that these treatments are often only successful in a subgroup of patients. One of the current challenges in oncology lies in identifying these patients and thereby avoiding treating patients with non-effective treatments and thus avoid side-effects of these drugs and reduce costs. This way, cancer medicine will become more and more personalized: a specific therapy for a specific tumor in a specific patient, with as little side-effects as possible. In this thesis, different ways of selecting patients are studied. Using PET scans with different radioactive labeled markers might be useful to predict whether a patient will benefit from a certain treatment. Furthermore, we studied whether biomarkers measured in blood samples can be used to predict side-effects in patients.