Abstract Lung adenocarcinoma (ADC) is a heterogeneous group of tumors associated with dramatically different survival rates, even when detected at an early stage. We hypothesized that a single cell proteomic approach would allow the dissection of cellular determinants of early lung ADC behavior. We developed a mass cytometry panel of 34 antibodies and validated their performance in four ADC cell lines (A459, H23, PC9 and H3211) and immune cells. We tested our panel in a set of 10 early stage lung ADCs, classified into long- (LPS) (n=4) and short-predicted survival (SPS) (n=6) based on radiomics features. Tumors were disaggregated and cryopreserved until mass cytometry analysis. We identified cellular subpopulations by clustering and analyzed differences in their distribution both within the tumor microenvironment and the epithelial compartment. To validate our results a tissue micro array was generated from lung tissue blocks from patients with LPS and SPS lung adenocarcinoma. Fluorescent staining was performed for PanCK, CD45, CD3, HLA-DR, DAPI. Cell nuclei were segmented using deep learning algorithm and were further processed in KNIME analytical platform where cell segmentation, feature extraction and cell classification were performed. The antibody panel captured the phenotypical differences in ADCs cell lines and PBMCs. When tumors were analyzed long-predicted survival tumors had a higher proportion of immune cells, whereas some short-predicted survival tumors had a higher proportion of fibroblasts/mesenchymal cells. Additionally, tumors show high degree of heterogeneity with distinct protein expression profiles among epithelial subpopulations, and some subsets with high HLA-DR expression were positively correlated with CD4+ and CD8+ T cells with LPS samples being enriched for such subsets. These results were further validated by Fluorescent staining on TMA slides. We found a positive correlation between HLA-DR expression on tumor cells and T cell number (r = 0.25, p = 2.2e-05). For this, in neighborhoods of 100 um diameter for each tumor cell, HLA-DR median signal intensity on neighboring tumor cells and number of T cells were calculated in Python and used as inputs for correlative analysis. Spatial analysis was performed in KNIME by calculation of distances from each T cell to nearest 1st and 2nd tumor cell, for which LPS tumors showed smaller distances for both 1st and 2nd tumor cell compared to SPS tumors (p = 0.039, p = 0.21). Our results demonstrate a distinct cellular profile of epithelial and stromal cells among indolent vs aggressive ADCs with higher HLA-DR expression in indolent tumors, which is associated with greater T cell infiltration. Our results illustrate the heterogeneity of T cell responses and HLA DR expression in lung adenocarcinoma and should further our understanding of mechanisms related to tumorigenesis. This work deserves further validation at the cellular and molecular level to gain further insights into tumor behavior. The work was supported by CA196405 to PPM. Citation Format: Maria-Fernanda Senosain, Tatiana Novitskaya, Georgii Vasiukov, Yong Zou, Aneri Balar, Deon B. Doxie, Jonathan M. Lehman, Rosana Eisenberg, Fabien Maldonado, Andries Zijlstra, Sergey V. Novitskiy, Jonathan M. Irish, Pierre P. Massion. Single cell proteomic analysis of lung adenocarcinoma identifies high HLA-DR expression to be associated with indolent tumor behavior [abstract]. In: Proceedings of the AACR Virtual Special Conference on Tumor Heterogeneity: From Single Cells to Clinical Impact; 2020 Sep 17-18. Philadelphia (PA): AACR; Cancer Res 2020;80(21 Suppl):Abstract nr PO-045.
Read full abstract