The development and testing of new treatments for cancer necessitates the use of models that accurately recapitulate drug response in human patients. In vivo systems have been found to be a poor proxy of in vitro behaviour and response. In part these differences are due to the lack of a tumour microenvironment and interactions with the immune and circulatory system. Increasingly xenograft models are important systems for investigating tumourigenesis and response to drug treatment, whereby cancer cells are implanted into nude mice either subcutaneously (under the skin) or orthotopically (organ of origin). Using RNA-seq, we have examined the transcriptional differences between HPAF-II cells in a series of different models that has revealed distinct differences not only between in vivo and in vitro models, but also between tumours implanted subcutaneously or orthotopically. Treatment using ETC-159, a novel inhibitor of PORCN that leads to a complete ablation of WNT signalling, reveals key differences in how these tumours respond depending on the model used in terms of the subsequent pathways that are up-regulated or down-regulated. Our findings highlight the importance of choosing the most realistic model of tumour growth and development when studying response to pharmacological inhibition.