Optimizing surveillance strategies in gastroenteropancreatic neuroendocrine tumours: Moving towards individualized management

Abstract

Important challenges in GEP NET management are the wide spectrum in clinical behaviour and heterogeneous course of disease. The aim of this thesis was to optimize and improve current surveillance strategies and promote a more tailored approach towards patients with GEPNETs. In the first part of this thesis, the current diagnostic armamentarium was evaluated. In the first two chapters we focused on non-functioning pancreatic NETs (pNETs) in MEN1 patients. We concluded in our systematic review with risk of bias assessment that current general biomarkers have important limitations in surveillance strategies for pNETs in patients with MEN1 and surveillance strategies are therefore depending on imaging. MRI and EUS are the most sensitive modalities and individualization of surveillance strategies is limited with only observed tumour growth providing restricted guidance. In chapter 3 we calculated the risk of developing a clinically relevant non-functioning pNET at several ages in MEN1, in order to optimize the age to start pNET screening. We also confirmed that further stratification based on demographic characteristics is not possible. Current guidelines support serial comparisons between images to evaluate changes in both sporadic and hereditary neuroendocrine tumours. Therefore we compared Choi criteria with RECIST v1.1 as definitions for disease status.We hypothesized that Choi criteria, using lower cut-off values for changes in tumour size and measures of tumour attenuation, could provide a more accurate reflection of disease status in hyper vascular and indolent tumours like GEPNETs. However, we concluded that RECIST had more clinical utility than Choi because of its prognostic value but that the RECIST approach is insufficient for NET disease assessment as well. We therefore explored the discriminative utility, predictive- and prognostic value of a novel molecular biomarker in GEPNETs in the second part of this thesis. In three chapters, we evaluated the NETest, a multigene blood analysis that could possibly delineate tumor behavior. The NETest uses gene expression of 51 marker genes, which is based on the quantity of circulating transcripts in the peripheral circulation. Gene expression is interpreted by four supervised machine learning algorithms to create an outcome that could reflect tumor activity. In a step wise validation, we illustrated that the NETest was not suitable as a screening test but did have an added value for prediction of PFS and disease recurrence in sporadic GEPNET patients. In the second part of this thesis, we developed a better understanding of the NEtest utilities and gained insight in the current test limitations and pitfalls. We concluded that accurate reflection of disease activity and therapeutic efficacy in GEPNETs are still requirements not reached but close this thesis with an extensive, yet hopeful discussion on future directions.