Key points are not available for this paper at this time.
Abstract The expression subtypes of lung adenocarcinoma (LUAD) capture tumors with distinct pathway activity, mutations and histopathology and also differentiate clinical outcomes. The microenvironments of these subtypes, proximal-inflammatory (PI), proximal-proliferative (PP), and terminal respiratory unit (TRU), have been described generally as immune hot, immune moderate and immune cold, respectively, but otherwise have not been analyzed at high resolution. Here, we aimed to characterize and compare tumor microenvironments between LUAD subtypes. Using spatially barcoded arrays and cDNA libraries (10x Genomics), we sequenced the spatial transcriptomes of a 6. 5mm2 plane of 14 LUAD tumors from the Applied Proteogenomics and Organizational Learning Outcomes (APOLLO) program. Spatial transcriptomes had a median of 3, 560 spots and a median of 4, 026 genes detected per spot. First, collapsing the spatial array to a bulk measurement per sample, we applied our published expression subtype predictor classifying 4 PI, 5 PP, and 5 TRU cases. We then decomposed each tumor’s spatial expression profile by unsupervised clustering, followed by signature scoring and collapsing into tumor, immune, and stroma tumor microenvironment (TME) components. Twelve of the fourteen tumors harbored multiple components while two tumors had one component. The region areas of TME components showed trends among the subtypes, with PI having the greatest immune area and PP having the greatest tumor area. Within each tumor, we calculated differentially-expressed genes between each TME component. Comparing TME genes to the subtype predictor genes, we found significant overlap (chi-square p 0. 001). This indicates that genes that are variable among bulk tumors also have variability within tumors. We then predicted expression subtype for decomposed compartments. Five tumors had the same expression subtype across their TME components, which we refer to as single subtype tumors. However, six tumors had more than one expression subtype prediction among the tumor’s TME components, which we call ‘multi-subtype tumors’. Multi-subtype tumors had lower bulk subtype prediction scores than single-subtype tumors (p 0. 01), indicating that the TME diversity among tumors affects the bulk expression subtype. Interestingly, the six multi-subtype tumors were in the PP and PI subtypes, suggesting greater TME component diversity than TRU. Calculating the spatial compactness of the tumors through continuity indices, we found that PI subtype trended with greater intermixing of TME components. In summary, the bulk LUAD expression subtypes capture differences between tumors and within tumors related to the tumor microenvironment. The views expressed in this abstract are solely of the authors and do not reflect the official policy of the Departments of Army/Navy/Air Force, Department of Defense, USUHS, HJF, or U. S. Government. Citation Format: Shaoqiu He, Camille Alba, Savannah Kounelis-Wuillaume, Teri J. Franks, Martin L. Doughty, Robert F. Browning, Craig D. Shriver, Clifton L. Dalgard, APOLLO Research Network, Matthew D. Wilkerson. Spatial decomposition of lung adenocarcinoma expression subtypes reveals tumor microenvironment characteristics abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 1142.
He et al. (Fri,) studied this question.