Livio Trusolino, M.D. Ph.D.
Translational models of cancer precision medicine Background: Advances in technologies have allowed the attainment of powerful insights into the molecular determinants of human tumours, and in many cases inactivation of individual oncogenic drivers results in tumor regression. However, this knowledge has been translated into effective treatments very slowly, due to difficulty in predicting how the complex mutational background and the adaptive resilience of cancer cells can influence the activity of the dominant oncogene and modify response to therapies.
Our projects aim at exploring the mechanisms of tumor dependency on oncogenic drivers and at understanding how such dependency is affected by genomic or functional modifiers, with an emphasis on colorectal cancer (CRC). Our experimental pipeline entails the deployment of multi-dimensional data for discovery and hypothesis validation, followed by cell-based mechanistic investigation and preclinical validation in patientderived tumorgrafts – ’xenopatients’. By this approach, we have contributed to finding that hyperactivation of the HER2 and MET oncogenes correlate with resistance to anti-EGFR therapies (cetuximab and panitumumab). To examine the global effects of somatic genetic changes in CRC on sensitivity to EGFR-targeted therapies, we performed complete exome sequence and copy number analyses of 129 tumors and analyzed their response to anti-EGFR blockade in xenopatients. We identified several alterations as potential mechanisms of resistance to this therapy as well as genetic lesions segregating in tumors with increased sensitivity to anti-EGFR therapy. Resistance to EGFR blockade could be overcome in xenopatients through combinatorial therapies targeting actionable genes. A typical response of CRC patients sensitive to EGFR neutralization is stable disease rather than massive regression. Mining of candidate gene expression outliers – coupled with retrospective analysis of patient material and in vitro functional studies – identified IGF2 as a predictor of poor sensitivity to cetuximab and as a new target. One limitation of xenopatients is that human tumour stroma is substituted by host (murine) components. However, this drawback can be exploited analytically to dissect the representation of human (cancer-cell specific) versus mouse (stromal-cell specific) traits in CRCs. By doing so, we contributed to re-categorising the transcriptional classification of CRC by demonstrating that a subtype displaying features of epithelial-to-mesenchymal transition and poor prognosis is not an expression of aggressive cancer cell dedifferentiation (as originally argued) but rather a read-out of abundant stromal content.
Conclusions and perspectives:
Our studies provide a systematic functional approach to evaluate response to targeted therapies in human cancer, highlight new mechanisms of responsiveness to anti-EGFR therapies, and provide a new vocabulary for the molecular management of colorectal cancer with immediate clinical implications.