Decoding spatial transcriptomics-based heterogeneity of TME in early-stage invasive pulmonary adenocarcinoma.

e20030 Background: Tumoral heterogeneity has been widely acknowledged to influence clinical outcome of lung adenocarcinoma (LUAD). This study investigated spatial gene profiling of different pathological subtypes of early-stage invasive LUAD with Stereo-seq transcriptomics technique. The tumor microenvironment (TME) was decoded with cellular and molecular events at perspectives of tumoral activities and immune activities. Methods: We collected spatial RNA profiles of nine Stereo-seq chips (captured area of 1 x1 cm) from 4 invasive LUAD patients. Pathologist identified histological subtypes including lepidic adenocarcinoma (LP), acinar adenocarcinoma (AC) and other invasive adenocarcinoma like papillary adenocarcinoma (IA) on adjacent H&E images. Spatial data matrix was obtained based on pathological annotation accordingly. Spatially resolved gene expression of each subtypes was obtain based on pathological annotation to conduct high resolution analyses. TME functional units were calculated by analysis of cellular neighborhoods (CNs) based on cell compositions within resolution of bin 100 (window of 50um x 50um). Sub-populations of tumoral cells were stratified by molecular functions (GSEA and GSVA score) and stemness (CytoTRACE). Results: All pathological subtypes shared the identical 10 heterogeneous CNs and 5 tumoral clusters (TM) with various propotions. Tumor cluster 0 (TM0) and TM3 were identified as the malignant population with significant high stemness and significant high activities of MTORC1 signaling, TGF-BETA signaling, PI3K-ATK signaling and hypoxia. AC and IA were mainly composed of TM0 and TM3. Compared to TM0, TM3 had higher EMT activity. We further identified two immune-related CNs (CN4 and CN8) with significant high GSVA score of Interferon-Alpha response, Interferon-Gamma response and Inflammatory response. Spatial association showed that the lymphoid-cells-dominated CN4 was enriched in LP and IA and nearly absent in AC. Meanwhile, the myeloid-cells-dominated CN8 was enriched in AC. GSEA score showed that immune activity of CN4 was more intense in IA when compared to LP. Notably, TM0 mainly present in CN8, and TM3 mainly present in CN4. Therefore, the LP was speculated to reach an inert and statical status. TME of AC was featured as colocalization of myeloid cells and tumor cells with high stemness. TME of IA was featured as colocalization of lymphoid cells and tumor cells with high EMT activity. Conclusions: This is the first study to delineate the homogeneity of tumor cells and heterogeneity of immune cells among different pathological types of NSCLC.