Abstract LB242: Prediction of patient response to neoadjuvant chemotherapy in breast cancer from their deconvolved tumor microenvironment transcriptome

Abstract

Abstract Introduction: The tumor microenvironment (TME) is a complex and dynamic ecosystem that plays critical roles in tumor development and clinical outcome. While the multifarious cellular interactions within the TME have been extensively studied with a significant emphasis on immunotherapy, understanding its role in chemotherapy outcome remains less explored. To this end, we present a novel generic computational framework named DECODEM (DEcoupling Cell-type-specific Outcomes using DEconvolution and Machine learning). DECODEM facilitates the exploration of the association of cell-type-specific gene expression profiles with the effectiveness of neoadjuvant chemotherapy (NAC) in treating breast cancer (BC). This approach holds promise for elucidating the nuanced ways in which the TME can modulate therapeutic efficacy, potentially paving the way for more personalized and effective cancer treatments. Methods: DECODEM employs cellular deconvolution of bulk transcriptomics to extract cell-type-specific gene expression profiles within the TME, and subsequently harnesses machine learning to construct cell-type-specific predictors of clinical outcomes. Utilizing DECODEM, we analyzed three publicly available cohorts of HER2-negative BC patients who received NAC. We assessed the predictive capabilities of the individual cell types as well as their collective influence, benchmarking against a state-of-the-art predictor that employs both clinical and transcriptomic features. We corroborated our findings in a single-cell (SC) transcriptomics cohort of patients with triple negative breast cancer (TNBC) undergoing chemotherapy. Further, we expanded DECODEM to DECODEMi, which accounts for cell-cell interactions (CCIs) among the predominant cell types, offering potential insights into the mechanistic processes governing clinical responses. Results: Our analysis indicated that the heterogeneous composition of the TME is crucial for chemotherapy response prediction in HER2-negative BC, beyond the influence of malignant cells alone. The key insights from our study are as follows:1. Gene expression of immune cells (myeloid, plasmablasts, B-cells) and stromal cells (endothelial, normal epithelial, cancer-associated fibroblasts) displayed comparable-to-superior predictive powers over both bulk expression and the state-of-the-art predictor.2. The predictive accuracy is further augmented by multi-cell-type ensembles, with a triad ensemble comprising endothelial, myeloid, and plasmablast cells achieving the highest predictive accuracy across multiple cohorts.3. DECODEMi can effectively capture the regulatory impact of CCIs (validated in a SC dataset sourced from the SC-TNBC cohort) and has identified novel CCIs likely to influence chemotherapy response in HER2-negative BC. Conclusion: We provide an unprecedented computational methodology to quantitatively assess the cell-type-specific influences of the TME on clinical outcome. Our analysis offers insights into the complex interactions among TME cells and their correlation with chemotherapy efficacy in HER2-negative BC, thereby enriching the knowledge base of BC treatment effectiveness. Citation Format: Saugato Rahman Dhruba, Sahil Sahni, Binbin Wang, Di Wu, Yael Schmidt, Eldad Shulman, Sanju Sinha, Stephen-John Sammut, Carlos Caldas, Kun Wang, Eytan Ruppin. Prediction of patient response to neoadjuvant chemotherapy in breast cancer from their deconvolved tumor microenvironment transcriptome [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr LB242.