Integrating tumor and stroma to understand and predict treatment response

点击翻译按钮获取中文摘要

ABSTRACT Through novel deconvolution approaches for bulk RNA sequencing analysis, we identified two tumor-intrinsic subtypes of PDAC (basal and classical) that we have confirmed, are robust, replicable, prognostic and predictive of treatment response. We found that the basal subtype is consistently associated with poor outcome and have shown, through analysis of two clinical trials, that patients with basal subtype tumors do not respond to the 1st-line therapy FOLFIRINOX. These results strongly support the idea that molecular subtypes may be used to select treatment. Given the impact of our tumor-intrinsic subtypes on therapy response, we developed a single sample classifier, PurIST, that is now a CLIA certified assay and being evaluated as an integral marker for treatment selection in a clinical trial. In parallel, we developed a de novo approach, DECODER to deconvolve bulk tumors into compartments that allows us to determine tumor and TME specific characteristics in patients. Using the deconvolution approaches that led us to identify tumor-intrinsic subtypes, we have found two types of PDAC stroma: activated, and normal where patients with activated stroma have shorter survival. We have shown that CAFs are the contributory cells in activated stroma. Patients can be found to have a combination of tumor/stroma subtypes and the combinations have different impacts on outcome, suggesting that it is critically important to understand tumor-stroma interactions and how they affect treatment response. Similarly, we find that i/myCAF may differentially educate basal vs. classical subtype lines Our findings provide strong support for our central hypothesis that CAFs and tumor cells have interactions that together may alter tumor progression and response, making it critical that we understand the heterogeneity of the stroma, and specifically CAFs, their interaction with the tumor, for tumor-stroma context specific treatment response. Our team is uniquely positioned to comprehensively characterize CAF/NAF-tumor heterogeneity and interactions, response to treatment, and develop an integrative CAF-tumor subtype classifier to predict treatment response of patients in standard of care, stroma and immune modulating trials.