Abstract SY18-02: Molecular genetic subtypes of pancreas and colon cancer: Phenotypes and potential therapeutic significance.

Abstract Pancreatic neuroendocrine tumors (PNETs) are relatively uncommon but nevertheless a serious, often fatal disease. While the prognosis of functional PNETs (secreting one or more hormones such as insulin at detectable levels in blood) is better, non-functional PNETs (NF; no hormones secreted) often present late with metastasis, and hence their prognosis is poor, with an overall 5-year survival rate of roughly 30%. Moreover, current standard therapies given to unselected NF PNET populations provide only modest and variable benefit. Seeking to stratify such PNET patients according to their likely responses to therapy, we have defined three molecular genetic subtypes (based on mRNA and microRNA transcriptome profiling) using human PNET samples. Interestingly, we observed that two of the three human PNET subtypes - ‘islet tumors’ (IT; insulinomas) and ‘metastasis-like primaries’ (MLP) - were similar at the molecular level to tumor subtypes found in a well-characterized genetically engineered mouse model of PNET. Both human and mouse PNETs of the MLP subtype had microRNA/mRNA transcriptome profiles similar to liver metastases and to precursor/stem cells. The distinctive characteristics of these subtypes may suggest new approaches for selective therapeutic targeting of human PNET, leveraging the capability to study subtype characteristics and drug responses in vivo using the mouse model that phenocopies two of the three PNET subtypes. Colorectal cancer (CRC) is a heterogeneous disease, and overall survival remains limited despite recent advances in the treatment of metastatic CRC. Unfortunately, the molecular classification methods currently available for CRC do not predict treatment strategies. Seeking new means of classification, we combined gene expression profiling with drug response data to develop clinically deployable assays that identify six integrated patient tumor subtypes with differential prognosis and cellular origin. We asked whether first line CRC therapies might exhibit selectivity for one or more of our subtypes. Indeed, we found that one of the CRC subtypes, which has better disease free prognosis (DFS) in the adjuvant setting, responds to cetuximab in the metastatic setting, whereas a second subtype with poor DFS responds to chemotherapy (but not cetixumab) both in the adjuvant and metastatic settings. As a proof of principle, we identified several subtype-specific CRC cell lines, and tested them as xenograft mouse models, which revealed similar drug responses to those of patients from the same subtype. Overall, the correlation of our CRC subtypes with patient outcome and drug responsiveness, along with clinically deployable means to identify them, support the proposition that targeting selected populations for treatment with this classification system could be clinically beneficial. Our informative cross-filtering between mouse and human tumors that revealed clinically significant subtypes in PNET (to be described) and PDAC (Collison, Sadananadam, et al Nature Medicine 2011) should motivate similar studies in genetically engineered mouse models of CRC, which may produce further knowledge about the biology and therapeutic sensitivities of these provocative new subtypes of CRC. Citation Format: Anguraj Sadanandam, Douglas Hanahan. Molecular genetic subtypes of pancreas and colon cancer: Phenotypes and potential therapeutic significance. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr SY18-02. doi:10.1158/1538-7445.AM2013-SY18-02