The prognosis of metastatic colorectal cancer (CRC) remains poor. Patients and physicians are in need of individual therapies and precise response predictions. We investigated the predictive capacity of primary tumour material for treatment response of metastases. Mutational landscapes of primary tumours and corresponding metastases of 10 CRC patients were compared. Cell line characteristics and chemosensitivity were investigated pairwise for primary and metastatic tumours of four patients. PDX models of one patient were treated in vivo for proof of concept. Driver mutations did not differ between primaries and metastases, while the latter accumulated additional mutations. In vitro chemosensitivity testing revealed no differences for responses to 5-FU and oxaliplatin between primary and metastatic cell lines. However, irinotecan response differed significantly: the majority of metastases-derived cell lines was less sensitive to irinotecan than their matching primary counterpart. Therapy recommendations based on these findings were compared to clinical treatment response and mostly in line with the predicted outcome. Therefore, primary tumour cell models seem to be a good tool for drug response testing and conclusion drawing for later metastases. With further data from tumour-derived cell models, such predictions could improve clinical treatment decisions, both recommending likely effective therapeutic options while excluding ineffective treatments.
Keywords: chemosensitivity; colorectal cancer; matching cell line pairs of primaries and metastases; metastasis; therapy prediction.