Clinical stratification of pancreatic neuroendocrine tumors by systematically integrating multimodal and clinical data

#3183

Introduction: Non-functional pancreatic neuroendocrine tumors (PanNETs) are heterogeneous with at least two transcriptome subtypes with differential biology, immune mechanisms and prognosis. However, it is challenging to understand how multimodal data interact with clinical variables and contribute to the disease heterogeneity and phenotypes.

Aim(s): To systematically integrate clinical variables with multimodal data (gene, microRNA and mutations) to better define clinically applicable non-functional PanNET subtypes that complement grade and improve patient stratification.

Materials and methods: We profiled 36 patient PanNET samples for mRNA, microRNA (miR) and mutations. We developed a novel machine-learning approach, (mPhenMap) multimodal phenotype mapping tool to integrate the profiles with 12 different clinical variables. We validated these findings using 173 RNAseq data.

Conference: 18th Annual ENETS Concerence (2021)

Presenting Author: Sadanandam A

Authors: Sadanandam A, Lawlor R, Mafficini A, Luchini C, Nyamundanda G,

Keywords: pancreatic neuroendocrine tumor, multimodal data integration, multiomics, clinical data integration, machine learning, artificial intelligence, subtypes, pancreatic cancer, phenotypes, PhenMap, mutations, gene expression, microRNA,