Abstract Background. Only a subset of non-small cell lung cancer (NSCLC) patients benefit from immune checkpoint inhibitors (ICIs). Thus, there is a clinical need to develop predictive biomarkers for ICIs. Deeper insights from the tumor microenvironment (TME) can lead to identification of spatial features associated with clinical responses to immune checkpoint inhibitors (ICI). In this study, we investigated functional profiles within cell subtypes as a surrogate for ICI clinical endpoints. Methods. We profiled a retrospective cohort of 39 NSCLC tissue cores from 27 patients treated with ICI, using highly multiplexed immunofluorescence (mIF) stained by the Phenocycler Fusion platform (Akoya Biosciences) capturing 45 proteins. Utilizing a deep learning based multiplex imaging analysis pipeline, cells were classified to 15 cell types by known marker expression and were further subclassified by unsupervised clustering. Cells were assigned to the tumor area or TME and more than 1000 spatial features were calculated based on cell type, marker positivity, and area assignments, and were compared between responders and non-responders to ICI therapy using Fisher’s exact and logrank tests. Results. Unsupervised cell subtyping of the 15 cell types identified 43 cell subsets, which were mostly segregated by their metabolism and activation status. Interestingly, all lymphocytes showed a similar pattern of clustering resulting in two clusters of metabolically active and inactive cells. The most significantly differentially expressed proteins between these cell types were oxidative phosphorylation proteins such as CS, SDHA and ATPA5, and other metabolic enzymes such as HK1, GLUT1 and LDHA. We found a direct connection between metabolic state, effector functions and tissue localization as the metabolically active lymphocytes exhibited higher levels of PD-1, MHC class I and II and CD44 positivity, and were more abundant within tumor infiltrating lymphocytes (TILs) and tertiary lymphoid structures (TLS). Unsupervised clustering of tumor cells demonstrated segregation to three main metabolic states - OXPHOS+, OXPHOS- and a third cluster (PPP+) which was characterized by upregulation of ASCT2, a glutamine transporter, as well as pNRF2 and G6PD, regulators of the pentose phosphate pathway. These cells also exhibited higher proliferation rate and CD44, a tumor stemness marker, positivity. All tumors with high content of PPP+ tumor cells (>40%) were resistant to PD-1 blockade (0/8 vs. 10/18 response rate in other tumors, p=0.009) and showed reduced overall survival (OS) rates (median OS of 27.6 vs 90.3 months, p=0.02). Conclusions. Taken together, this study reveals connections between metabolic states, effector functions, and immunotherapy outcome and contributes to the evolving landscape of predictive biomarkers for ICI therapies. Citation Format: James Monkman, Rotem Czertok, Shai Bookstein, Becky Arbiv, Yuval Shachaf, Ron Elran, Kenneth Bloom, Oscar Puig, Ken O'Byrne, Ettai Markovits, Arutha Kulasinghe. Spatially resolved cell profiling unveils tumor metabolic states associated with immunotherapy response in NSCLC [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(6_Suppl):Abstract nr 1158.
Read full abstract