155 MHC class I antigen presentation is associated with an inflamed SCLC tumor microenvironment characterized by a higher density of cytotoxic T-cells in closer proximity to tumor cells

BackgroundSmall cell lung cancer (SCLC) is generally known to exclude immune cells and durable responses to immunotherapies are rare. Only very few biomarkers to inform immuno-oncology (IO) treatments are established in clinical practice thus far. Recently, four major SCLC subtypes (SCLC-A, SCLC-N, SCLC-P and SCLC-I) were described. Whereas the first three are characterized by activation of specific transcription factors, the SCLC-I (inflamed) subtype is characterized by an inflamed gene signature, high expression of MHC class I (MHC-I) antigen presentation and shows the greatest benefit from addition of immunotherapy to chemotherapy treatment [1,2]. Importantly, MHC-I is epigenetically silenced in the vast majority of SCLC and the presence of MHC-I could serve as a biomarker for the identification of SCLC-I cases. [2]. Here, we aimed to assess the biology of MHC-I high SCLC cases to investigate its role as a biomarker to inform cancer immunotherapies.MethodsWe combined the power of artificial intelligence (AI)-driven computational pathology with multiplex immunofluorescence (mIF) to gain critical insight into the tumor microenvironment (TME) of SCLC. 125 SCLC formalin-fixed, paraffin-embedded tissue samples were stained with a mIF panel consisting of six markers: PanCK, CD8, CD68, PD-1, PD-L1, and Ki67. We assessed the phenotype and spatial location of each cell in the pathologist-annotated tumor center and within the AI-segmented stroma and tumor epithelium. Pathologists classified immunohistochemically stained MHC-I slides from the same tissue blocks as MHC-I high, medium, or low according to their H-scores (low: ≤30; medium: 31–139; high: ≥140). TME characteristics between those groups were compared.ResultsIn all measured regions, we found higher densities of CD8+ and particularly PD-1/CD8 double positive T-cells in the MHC-I high group. Moreover, we observed the highest proportion of PD-1 positivity among cytotoxic T-cells in the tumor epithelium of MHC-I high samples, which also showed a high density of PD-L1+ tumor cells. Average distance of PD-1+ T-cells to their nearest tumor cell was lowest in the MHC-I high group. In the same group, an average of 19.3% of tumor cells in the epithelium had at least one PD-1+ T-cell within a 50 µm radius, while in the MHC-I low group this average was only 8.9%.ConclusionsWe utilized cutting-edge computational pathology to establish MHC-I as orchestrator of the composition and spatial arrangement of an inflamed SCL