ABSTRACT Although studies about tumor-infiltrated leukocytes (TILs) have attracted substantial attention to understand tumor microenvironment and related immune response, considerable methodological gaps remain for evaluating intratumor heterogeneity of TILs in multi-region omics data. The proposed study is directly motivated by our collaborations with lung cancer medical oncologists in the investigation of intratumor heterogeneity and lung cancer patients' treatment outcome. The primary objective of this proposal is to develop accurate statistical models to quantify TILs by combining multi-region omics data and the prior knowledge about leukocytes. In this project, (Aim 1) we propose a Bayesian modeling approach to estimate intratumor heterogeneity of TILs from multi-region transcriptomics data. The model overcomes the limitations in existing works and specifically addresses the correlations within the same tumor and the variability at the patients' level. We will further generalize the approaches to account for different data distributions to address the estimations in the multi-omics setting (Aim 2). We will apply the proposed methods to the MD Anderson Cancer Center Intra-Tumor Heterogeneity (MDACC-ITH) project for lung cancer patients and the TCGA datasets. From an application perspective, our proposed methods of maximizing the use of existing multi-region omics data and incorporating complex data structure is cost-effective and may directly improve our understanding of TILs and their relationship with patient outcomes. Although motivated by lung cancer research, the statistical methods will be useful for estimating intratumor heterogeneity of TILs in other cancer types. All software for statistical tools developed in this project, once validated, will be made available to the broader research community.

NARRATIVE/PUBLIC HEALTH RELEVANCE Intratumor heterogeneity of tumor-infiltrated leukocytes (TILs) is an important phenomenon of cancer biology with potentially profound clinical impacts. Multi-region omics data provide a promising direction to understand TILs and their intratumor heterogeneity. This project will produce novel statistical methods for accurately quantify TILs from multi-region omics data, which help understanding how intratumor heterogeneity of TILs relate with patient outcome.