8133 Spatial Transcriptomic Profiling of Pituitary Corticotroph Carcinoma: Uncovering Architectural Complexity

Abstract Disclosure: D. Zhang: None. M. Bergsneider: None. M.B. Wang: None. W. Kim: None. H. Vinters: None. A.P. Heaney: None. Most pituitary corticotroph adenomas are benign but ∼15% are refractory to conventional surgical, medical and even radiotherapeutic management. This tumor subset invade sellar adjacent structures, and in rare cases metastasize locally within the central nervous system or to neck lymph nodes (LN) or more distant organs such as the liver. To better understand the potential interactions between tumor cells and neighboring tissues during these progressive stages, we used spatial transcriptomics (ST, Visium, 10x Genomics) to profile the genetic composition and native architecture of 4 surgically resected corticotroph tumor paraffin-embedded samples collected from the same patient over several years. Samples 1& 2 invaded the sino-nasal mucosa, sample 3 the parietal dura, sample 4 the clival bone, and sample 5 had metastasized locally to a neck LN. Quality of the 4 samples was confirmed by DV200 score (51 ± 7%). An average of 2,473 tissue spots per tissue sample were obtained after sequencing with a median of 4,893 genes detected per spot. We then used the Seurat analytic pipeline to reconstitute spatial transcriptional distribution that authentically mirrored histologically verifiable structures. These included the main tumor area itself, it’s invading tumor edge, adjacent normal epithelium, dura, and LN tissue. Further deconvolution analysis revealed a distinct cellular composition at the invading tumor edge which was tissue context specific. For instance, a proliferative cell population was noted between the central tumor area and it’s invasive margin, B cell and fibroblast cell enrichment was prominent at sino-nasal mucosal invasive margins, whereas endothelial and pericyte cells were prominent along the tumor/ normal pituitary border. . As expected, the central tumor area expressed genes associated with corticotroph hormone production including POMC, NNAT, GAL and GNAS, genes associated with mitochondrial energy metabolism such as ATP5F1E, NDUFA8 and COX7B as well as transcripts related to lipid kinase activity (GKG, CERKL, PIK3C2G, DGKK). Most intriguingly, tumor cells at the fore-front of the invading edge expressed high level of cyclin-dependent kinases, potentially reflecting higher proliferative rates of these invasive margin cells. Additional studies to validate and define the role of theses invasive margin transcripts are now warranted to translate our profiling findings into potential drug targets. In summary, this is the first use of spatial transcriptomics to delineate the evolving transcriptomic features in a series of surgically resected pituitary corticotroph tumors from the same patient that progressed over time to a locally metastatic corticotroph carcinoma. It provides unparalleled insights into the differential transcriptional topography of corticotroph tumor tissue and most importantly, its invading edge and may identify novel therapeutic targets. Presentation: 6/3/2024