Multiregional profiling revealed intra-tumor transcriptomic heterogeneity associated with the prognosis in non-small cell lung cancer

Abstract Introduction: Intratumor heterogeneity (ITH) describes the distinct tumor cell populations and microenvironments within the same tumor, which may profoundly impact cancer evolution and clinical outcomes. Non-small cell lung cancer (NSCLC) is not only a genetically diverse disease but also has high transcriptomic heterogeneity (RNA-ITH). The RNA-ITH limits the reproducibility of expression-based prognostic models, which is poorly understood. Methods: To address the issue, we investigated the effect of RNA-ITH on prognosis at both gene and signature levels using multiregional RNA-seq data from 45 NSCLC patients (145 regions) in the TRACERx study. We also performed multiregional RNA-seq of 25 NSCLC tumors (64 regions) for independent validation. Results: At the gene level, we found that the maximal expression of hazardous genes (Hazard Ration (HR) > 1) and the minimal expression of protective (HR < 1) genes across different regions within a tumor are more prognostic than their average expression. As for prognostic signatures, we first designed five different functions to transform the multiregional expression of signature genes into patient-level values. To calculate individual risk scores, we applied them to assist two existing prognostic gene signatures, ORACLE (Outcome Risk Associated Clonal Lung Expression) and WTGS (whole-transcriptomic gene signature). As a result, the best performance was achieved using the combination of maximal hazardous signature expressions and minimal protective signature expressions. We next developed a new signature called PACEG (Prognosis-Associated Clonally Expressed Genes) and proposed a multiregional assay for higher prognostic accuracy in NSCLC. We demonstrated significant improvement in PACEG performance by leveraging RNA-ITH captured by multiregional expression of signature genes. Finally, we utilized the same strategy to study the impact of tumor immune microenvironment ITH on patient prognosis. Consistently, the minimal/maximal infiltration of protective/hazardous immune cells across tumor regions was the best measurement associated with prognosis in NSCLC. These results were independently validated by our local datasets. Conclusions: The prognosis of NSCLC patients is often driven by the most aggressive tumor subclones. Our study proposed a novel strategy to incorporate RNA-ITH with expression-based prognostic models. Multiple distinct tumor regions should be considered to overcome the ITH issue for better prognostic evaluation, e.g., using the minimal/maximal expression of protective/hazardous signature genes across all regions to calculate the risk score in individuals. We also developed the PACEG panel composed of 26 genes that could be potentially applied in clinical specimens to identify high-risk NSCLC patients who may benefit from intensified adjuvant therapy. Citation Format: Chenyang Li, Thinh T. Nguyen, Jian-Rong Li, Xingzhi Song, Ignacio I. Wistuba, Andy Futureal, Jianhua Zhang, Shawna M. Hubert, Jia Wu, Jianjun Zhang, Chao Cheng. Multiregional profiling revealed intra-tumor transcriptomic heterogeneity associated with the prognosis in non-small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 97.