Outcome Driven Persona-Typing For Precision Oncology: Beyond A Genomics Centered View Of Individualized Therapy.

4130 Background: Precision oncology is currently being defined mainly by a genomics-oriented view to tumor biology. However, a multi-omic view to tumor biology and more accurate outcome prediction is emerging. Combining this with treatment history, clinical-epidemiological data, and outcomes data may provide patient-specific descriptors that in N-dimensional space constitute population based “personas” that share common outcome destinies and identify response predictors to any given therapy. Methods: We utilized our database from 919 pancreatic adenocarcinoma patients within our ongoing registry study as a feasibility study. Commercial exome (315 genes by NGS) and proteomic data (24 proteins by IHC) as well as previous and current treatment history, epidemiological data and outcomes data were collected on every patient. Overall Survival (OS) was calculated and 10 individual outcome “personas” were created that spanned short-term survival ( < 6 months) to long term survival (32-110 months). Statistical analysis of individualized gene, protein, specific treatment type, disease stage, location, age, gender, ethnicity, was used to determine key principal components that significantly (p < 0.05) described each outcome persona to create a unique persona-type identifier. Results: Proteomic information was significantly associated with outcomes more frequently then genomic information (p = 0.02). Longest term outcome personas (OS > 32 months) were characterized by increased PD1 and decreased TS protein levels along with increased frequency of BRCA2 genomic alterations and treatment with off-label targeted therapy. Shorter term personas (OS < 6 months) were described by high TS protein levels along with genetic alterations in MYCL1, MYST3, VEGFA, ZNF703 and KEL. Conclusions: Persona-typing can be used to define and map key characteristics that associate with outcome and specific treatment. In the future, individual patients can be mapped to pre-defined personas that could more accurately describe outcome destiny and optimized/personalized therapy options.