Activation of natural killer (NK) cells with the cytokines interleukin-12 (IL-12), IL-15, and IL-18 induces their differentiation into memory-like (ML) NK cells; however, the underlying epigenetic and transcriptional mechanisms are unclear. By combining ATAC-seq, CITE-seq, and functional analyses, we discovered that IL-12/15/18 activation results in two main human NK fates: reprogramming into enriched memory-like (eML) NK cells or priming into effector conventional NK (effcNK) cells. eML NK cells had distinct transcriptional and epigenetic profiles and enhanced function, whereas effcNK cells resembled cytokine-primed cNK cells. Two transcriptionally discrete subsets of eML NK cells were also identified, eML-1 and eML-2, primarily arising from CD56bright or CD56dim mature NK cell subsets, respectively. Furthermore, these eML subsets were evident weeks after transfer of IL-12/15/18-activated NK cells into patients with cancer. Our findings demonstrate that NK cell activation with IL-12/15/18 results in previously unappreciated diverse cellular fates and identifies new strategies to enhance NK therapies.

People diagnosed with cancer and their formal and informal caregivers are increasingly faced with a deluge of complex information, thanks to rapid advancements in the type and volume of diagnostic, prognostic, and treatment data. This commentary discusses the opportunities and challenges that the society faces as we integrate large volumes of data into regular cancer care.

This talk presents a hybrid computational and experimental strategy to uncover interactions between neoplastic cells and the microenvironment during pancreatic carcinogenesis. As pancreatic cancer develops, it forms a complex microenvironment of multiple interacting cells. The microenvironment of advanced pancreatic cancer includes a dense composition of cells, such as macrophages and fibroblasts, that are associated with immunosuppression. New single-cell and spatial molecular profiling technologies enable unprecedented characterization of the cellular and molecular composition of the microenvironment. These technologies provide the potential to identify candidate therapeutics to intercept immunosuppression in pancreatic cancer. Inventing new mathematical approaches in computational biology are essential to uncover mechanistic insights from high-throughput data for these precision interception strategies. Here, we demonstrate how converging technology development, machine learning, and mathematical modeling can relate the pancreatic precancer microenvironment to carcinogenesis and therapeutic response. Combining genomics with mathematical modeling provides a forecast system that can yield computational predictions to anticipate when and how the cancer is progressing for therapeutic selection. This mathematical forecast system will empower a new predictive oncology paradigm, which selects therapeutics to intercept the pathways that would otherwise cause future cancer progression. Citation Format: Elana Judith Fertig. Forecasting pancreatic carcinogenesis through spatial multi-omics. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr PL03-04.

Spatial molecular profiling has the potential to provide us with crucial data for revealing the principles of tissue architecture. Using this technology we can create atlases with comprehensive measurements on the layers of gene activity—RNA, protein, small molecules, etc.—that can then be explored for discovery. The previous major step in molecular atlasing came with the advent of single-cell RNA-seq, which allowed us to delineate cell populations corresponding to cell types and cell states. However, this technology is unable to also reveal the spatial context for insights on the function of these populations, a problem now solved with spatial profiling. Though it is scientifically intoxicating to create spatial profiling maps, it is important to recognize the limits of map-making, with the adage that “the map is not the territory” reminding us that maps must be constantly tested against reality. More importantly, we must resist the temptation to make a perfect scientific map as our goal—which at its logical conclusion, such a map would be a ridiculous point-by-point replication of the original—a scenario famously immortalized by Jorge Luis Borges’ short story “On Exactitude in Science.” Instead, we wish to learn the principles underlying tissue architecture. Spatial profiling may lead us to learn, for example, how a “gene’s-eye” perspective may be better suited to explain tissue architecture in comparison with a “cell’s-eye” perspective. We may also learn principles of how complex tissue organizations may be the manifestation of a simpler set of underlying relationships. Spatial profiling promises to advance our understanding by revealing such principles, the search for which has been an enduring goal of biology. Spatial multiomics technologies give us unprecedented access to in situ cell-to-cell interactions. With it, we can now shift focus from identifying the individual building blocks of cancer growth and progression to understanding what cellular interactions signal the onset of disease. Currently, precision medicine approaches select therapies based on the pathways that artificial intelligence methods associate with malignancy at the time of high-throughput data measurement. This static, cell- and gene-centric paradigm disregards the complex adaptation of a tumor’s molecular and cellular pathways during carcinogenesis and treatment, resulting in tumor progression and therapeutic resistance. As a chaos theorist, I researched a method for weather prediction called data assimilation, which estimates atmospheric variables over time between gaps of measurements. I hypothesize that spatial multiomics technologies will enable the power of these computational systems to develop comparable forecasting techniques for molecular and cellular states of biological systems, such as tumors. Developing this requires techniques that can define the variables regulating biology from the complex, multi-scale systems containing thousands of genes and proteins in hundreds of cell types in tissues. Moving forward, I see technologies focused on spatiotemporal measurement on the horizon. With these data, scientists will be closer than ever to identifying the individual variables needed to apply data assimilation methods to cancer research. Doing this will identify the molecular and cellular components causing tumor evolution, enabling the next generation of predictive medicine for cancer and all human disease. Spatial molecular properties of tissues encompass the distribution, localization, and interactions of specific molecules within a 3D context. These properties offer valuable insights into tissue organization, function, and regulatory processes. Exploring the spatial molecular properties of tissues enables the discovery of critical biological information, including cellular interactions, signaling pathways, and disease-related molecular alterations. From my perspective, the greatest potential of spatial molecular profiling lies in its ability to comprehensively understand the 3D spatial molecular properties of tissues. Two recent studies employed spatial transcriptomics and multiplexed immunofluorescence imaging techniques to construct 3D tissue maps in cases of colon cancer and lung cancer. These studies identified large and interconnected 3D structures, such as tumor buds and tertiary lymphoid structures, which cannot be fully characterized using 2D tissue sections at a local level. It is fascinating to observe how spatial molecular profiling has contributed to uncovering previously unrecognized tissue properties in 3D. However, the high cost associated with spatial molecular profiling technologies poses a challenge when it comes to constructing 3D tissue maps for large population-based studies. One potential solution is to computationally predict spatial molecular information using readily obtainable and cost-effective histopathology images. A promising direction for future research involves developing innovative computational algorithms capable of reconstructing 3D spatial molecular tissue maps from histopathology images. Successful implementation of such methods would allow for the study of the 3D spatial molecular properties of tissues in biobank samples, enabling the correlation of these properties with clinical outcomes and prognosis. Spatial molecular profiling methods are powerful tools to study how cells, the central units of our tissues and organs, function and communicate within their local microenvironments. This has been made possible through innovative genomics, transcriptomics, epigenomics, and targeted proteomics-based methods that have shed light on the spatial and molecular organization of cells, elucidating crucial developmental and disease-related processes. In the upcoming years, we can anticipate a further technology boom that will advance all spatial omics disciplines (DNA, RNA, protein, and metabolite level). This includes the development of new technological and computational pipelines for spatial multiomics concepts that integrate different information layers from the same biological sample. A particularly exciting development is the rise of ultra-sensitive mass-spectrometry-based spatial proteomics. This approach, while still in its infancy, has the potential to profile thousands of proteins from few, or even single cells, complementing spatial transcriptomics and targeted proteomics data. The unique power of spatial proteomics lies in its ability to decode the quantitative, phenotype-linked proteome, illuminating distinct cell identities and functions. For the first time, we could address how the proteome is influenced by the type and state of neighboring cells and learn how diverse cell interactions are linked to complex disease outcomes and therapeutic responses. This could pave the way toward more accurate diagnoses and highly potent targeted therapies in diseases such as cancer and inflammatory and neurodegenerative disorders. After more than 200 years, we are finally rediscovering one of the most important fields in biology and medicine at the omics scale: histology, where, long ago, “crude” stains enabled the discovery of cell types like pyramidal cells, the causes of diseases such as bacteria causing leprosy, and functional tissue entities like the nephron. Today, we have the capability to profile high-dimensional molecular states at the resolution of single molecules in native tissue sections and with relatively high throughput. This opens up enormous potential for discovery once again. We now have access to information-rich, structured data that are just waiting to be mined and turned into insights. But what can we actually hope to discover? It seems that the fundamental breakthroughs have already been achieved in the last 200 years. However, with these data of unprecedented quality becoming available, we can now tackle the next level: uncovering spatially variable molecular cell states, understanding complex disease causes involving cross-cell molecular circuits and unraveling tissue functions that emerge from the spatio-molecular context. Wouldn’t it be amazing if a small tissue biopsy could reveal the molecular basis of a patient’s undiagnosed tissue-disruptive disease, particularly those caused by rare or complex immunological defects? The comprehensive assessment of molecular states and interactions, within their native spatial context, might just hold the key. Biological tissue performs complex functions through spatial organization and interactions among many cells. Spatial transcriptomic (ST) technologies offer an unprecedented opportunity to see into the mechanisms behind cellular organization and communication. Tissue function depends not only on what cell types are present and in what proportion but also on their spatial arrangement. Do groups of cells arrange themselves in clumps, salt-and-pepper mixtures, layers, or even region-within-region hierarchical structures? And how does this arrangement differ between diseased and healthy tissue? Spatiotemporal reconstruction of static ST data (e.g., via deep-learning-based SpaceFlow analysis) allows identification of novel patterns of cells to address some of these questions. Cellular communication responsible for such organization, which often involves diffusive or contact-based ligand-receptor interactions, can also be predicted using ST data via biophysical modeling (e.g., CellChat) or machine-learning methods (e.g., COMMOT), although in-depth analysis requires further advances in method development and data-collection techniques. When a disease disrupts spatial structure, what communication pathways cause changes or are changed themselves? Cells continually receive an inflow of signals from other cells, which are processed through intracellular gene regulations, and may cause in turn an outflow of signals that affect other cells. These flows of cause-and-effect relations, which are spatiotemporal in nature and drive every biological process, are often instigators of diseases. For the first time, such challenging questions of interplay between intercellular and intracellular communication, without even the need to pick and choose a specific molecule in advance, may be answered unbiasedly at a system level using ST data. With spatial transcriptomics, we can now study tissue heterogeneity while preserving its spatial context. Most existing analytical methods have focused on the analysis of single-tissue slide with some recent tools aiming to align and integrate consecutive tissue sections. However, much less has been done on comparative spatial omics analysis to spot the differences between phenotypic groups such as normal vs. cancer, mutant vs. wild type, etc. Such differences could be tissue niches, cell hubs, or gene modules specific to certain phenotype. Over the past year, spatial technologies have been expanding the scope to encompass spatial multiomics, enabling simultaneous measurement of multiple omics (e.g., protein and RNA, chromatin accessibility and RNA) on the same tissue slice while retaining the spatial context. To exploit the richness of the data generated by such techniques, analytical methods need to accomplish two inter-related tasks, integrating multiple data modalities (i.e., multiple omics) and integrating omics data with spatial information. However, most currently available algorithms target either spatial single omics or non-spatial multiomics data. Besides multiomics, spatial technology has also advanced rapidly to increase its spatial resolution and gene coverage with the latest development such as CosMx, Stereo-seq, and MERSCOPE, offering subcellular resolution and covering thousands of genes. Advanced image processing and computer vision techniques are needed for precise cell segmentation and cellular compartment separation to facilitate downstream analyses such as spatial organization of mRNA or protein molecules within the cell and cell-cell interactions. The data size of spatial omics is also increasing, which has also urged algorithm developers to take into consideration computing time and memory usage. Maps inform us about the directions, connections, and centers during travel. Cellular maps contain molecular trajectories, interactions, and hubs throughout state transitions and dynamic cellular decision-making processes. Current spatial omics methods yield snapshots of the cell’s position and molecular constituents in tissue maps. While such information is static, we might be able to leverage emerging tools in systems biology and deep learning to extract the history and future of a cell’s state. Would the cell neighborhoods be predictive of the past states of a cell, or can one tell what the next decision of a cell will be by looking at the current spatial niche of that cell? The expectation is to expand our capabilities of “pseudo” time analysis to spatial molecular data, enabling dynamic modeling of cells in tissues. Molecular maps are wired hierarchically in tissues. Organelle communication, phase separation, subcellular organization, extracellular distributions, cell-cell junctions, multi-cellular emergent features, and tissue anatomy make up the bottom-up structural organization of a biological specimen. The integrated, multimodal, and cross-scale analysis of these spatial features in health and disease will be crucial. Can we predict an early disease manifestation of morphological and molecular defects in these highly connected spatial networks? Translational utilization of such spatial diagnostic assays will possibly increase the quality of life in our biotechnology-driven society. Finally, spatial molecular profiling might be used to quantify and model the subcellular architecture of non-invasively collected blood cells complementary to tissue maps from the same individual. The hope is a predictive framework to approximate the quasi-tissue state in blood signatures. M.L. receives research funding from Biogen Inc. unrelated to the current manuscript. E.J.F. is on the scientific advisory board of Viosera Therapeutics and is a paid consultant for Merck and Mestag Therapeutics. J.K. is an inventor on patent application number 17156392 entitled “Molecular spatial mapping of metastatic tumor microenvironment.”

Summary of sequence analysis of adenoid cystic carcinoma.

Triple-negative breast cancer (TNBC) is an aggressive subtype associated with early metastatic recurrence and worse patient outcomes. TNBC tumors express molecular markers of the epithelial-mesenchymal transition (EMT), but its requirement during spontaneous TNBC metastasis in vivo remains incompletely understood. We demonstrated that spontaneous TNBC tumors from a genetically engineered mouse model (GEMM), multiple patient-derived xenografts, and archival patient samples exhibited large populations in vivo of hybrid E/M cells that lead invasion ex vivo while expressing both epithelial and mesenchymal characteristics. The mesenchymal marker vimentin promoted invasion and repressed metastatic outgrowth. We next tested the requirement for five EMT transcription factors and observed distinct patterns of utilization during invasion and colony formation. These differences suggested a sequential activation of multiple EMT molecular programs during the metastatic cascade. Consistent with this model, our longitudinal single-cell RNA analysis detected three different EMT-related molecular patterns. We observed cancer cells progressing from epithelial to hybrid E/M and strongly mesenchymal patterns during invasion and from epithelial to a hybrid E/M pattern during colony formation. We next investigated the relative epithelial versus mesenchymal state of cancer cells in both GEMM and patient metastases. In both contexts, we observed heterogeneity between and within metastases in the same individual. We observed a complex spectrum of epithelial, hybrid E/M, and mesenchymal cell states within metastases, suggesting that there are multiple successful molecular strategies for distant organ colonization. Together, our results demonstrate an important and complex role for EMT programs during TNBC metastasis.

Background: Rhabdomyosarcoma (RMS) is the most common soft-tissue sarcoma of childhood, and RAS pathway mutations are the known driver mutations in the majority of fusion-negative (FN) RMS. Recent studies have demonstrated that HRAS mutations are enriched in infant cases of FN-RMS and can be associated with an aggressive clinical course and inferior outcomes. Using HRAS-mutant RMS cell lines and xenograft models, we have demonstrated that tipifarnib (farnesyl transferase inhibitor, FTI) decreases ERK signaling, decreases in vitro proliferation, and decreases in vivo tumor growth. The effects of tipifarnib can be incomplete, however, leading only to partial or short-lived responses. Limitations may be due to adaptive or acquired resistance, suggesting that HRAS-mutated FN-RMS may be sensitive to pathway inhibition with combination therapy that prevents or delays the emergence of adaptive resistance. Trametinib (MEKi) inhibits tumor growth in xenograft models of FN-RMS but has only modest activity as a single agent, potentially due to the release of negative feedback and activation of upstream signaling. The efficacy of inhibition with FTI and MEKi has not been previously explored in RAS-driven FN-RMS. Materials and Methods: We examined the transcriptional effects of tipifarnib in FN-RMS cell lines using bulk RNA-sequencing to identify therapeutic vulnerabilities that may be exploited by RAS-directed therapies. Additionally, we utilized in vitro cellular proliferation assays, soft agar colony-forming assays, and immunoblot to evaluate the effects of tipifarnib in combination with trametinib on cell growth, differentiation, and signaling via RAS effector pathways. Results: Analysis of RNA sequencing data revealed downregulation of ERK transcriptional output genes upon treatment with tipifarnib, confirming the critical role of the MEK-ERK pathway in mediating the response to farnesyltransferase inhibition. We, therefore, tested tipifarnib in combination with trametinib in HRAS-mutant FN-RMS cell lines and observed additive dose-dependent 2D and 3D growth inhibition in response to the combination. In HRAS-mutant cells, tipifarnib and trametinib more potently reduced ERK phosphorylation than either drug individually, indicating effective RAS pathway inhibition. Additionally, we found that the combination of tipifarnib and trametinib induced myosin heavy chain expression in HRAS-mutated cell lines, suggesting both inhibition of proliferation and promotion of myogenic differentiation. Conclusions: Our data suggest that the combination of the FTI tipifarnib and the MEK inhibitor trametinib is active in models of HRAS-driven FN-RMS and may represent an effective therapeutic strategy for a genomically-defined subset of patients with FN-RMS. No conflict of interest.

In murine models of breast cancer the histone deacetylase inhibitor entinostat increases CD8+ effector: FoxP3+ regulatory T-cell ratios (CD8/FoxP3), and improves the efficacy of immune checkpoint inhibitors. We identified a recommended phase II dose (RP2D) for the combination of entinostat, nivolumab and ipilimumab (ETCTN-9884, manuscript submitted). We report combined safety and efficacy results of participants with HER2-negative breast cancer treated in dose escalation (n=6) and expansion cohorts treated at the RP2D (n=18). Participants received entinostat PO 5mg weekly x 2 (run-in), then 3-5mg weekly entinostat PO, 3mg/kg q2 weeks nivolumab, and 1mg/kg q6 weeks ipilimumab IV (max 4 doses ipi, RP2D). Primary endpoint: Safety (CTCAE v5). Secondary endpoints: Change in tumor CD8/FoxP3 ratio (integrated biomarker); Objective response rate (ORR). We obtained tissue samples at baseline, after 2 week run-in, and after 8 weeks of combination therapy and completed immunohistochemical staining for CD8, FoxP3 and PD-L1. Blood samples at each timepoint were obtained and Luminex analysis for global changes in plasma cytokine expression was performed. Amongst 24 participants [12 hormone receptor-positive (HR+), 12 triple-negative (TNBC)], median age was 54.5 years (range 38-77) and median prior therapies 6.5 (range 1-13). Median cycles received was 2 (range 1-17). Grade 3/4 AEs included anemia (N=4, 17%), decreased neutrophil count (N=3, 13%), and increased lipase (N=2, 8%). Most common immune-related (ir) AEs included rash (N =7, 29%), hypothyroidism (N =5, 21%), and pneumonitis (N=2, 8%). ORR by RECIST (v1.1) was 30% (6/20 evaluable), and by irRECIST was 20% (4/20 evaluable), including a complete response in a participant with TNBC. The combination of entinostat, nivolumab and ipilimumab at the RP2D was associated with expected (ir) AEs in advanced HER2-negative breast cancer. An ORR of 30% suggests this combination should be evaluated further. Correlative analyses from serial biospecimens pre- and post-therapy to evaluate the immune response and landscape will be presented.

Hepatocellular carcinoma (HCC) is a major global cause of mortality. The epithelial-mesenchymal transition (EMT) transcription factor TWIST1 has been implicated in the invasion and metastasis of HCC, but the mechanism is unclear. Integrated transcriptomic and proteomic analyses on primary liver tumors and lung metastases from a spontaneous Twist1-dependent metastasis mouse model of MYC-induced HCC identified the hexosamine biosynthetic pathway (HBP) as a cancer cell autonomous mechanism for EMT-mediated invasion and metastasis, which was further validated in vitro and ex vivo. RNA-sequencing and mass spectrometry-based proteomics were carried out on primary liver tumors, lung metastases, normal livers and normal lungs from a Twist1-dependent metastasis mouse model of MYC-induced HCC (LMT); a metastatic MYC-induced HCC mouse model dependent on Twist1 phosphorylation mimetic mutant (LMT-DQD); a non-metastatic mouse model of MYC-induced HCC (LM); a non-metastatic MYC-induced HCC mouse model with the Twist box mutant (LMT-F191G); and wildtype mice. Data was analyzed in R. Organoids were derived from LM liver tumors, embedded into collagen and imaged using time-lapse DIC microscopy. Cell lines were derived from LM and LMT liver tumors, and along with human HCC cell lines Huh7, Hep3B and HepG2, were utilized for in vitro migration and invasion assays. 91 genes were differentially expressed between lung metastases versus primary liver tumors; primary liver tumors versus normal liver; and normal lung versus normal liver with p<0.05. Gene Ontology and KEGG Pathway Enrichment analyses revealed an enrichment of metabolic processes. Gfpt2, the gene encoding for the rate-limiting enzyme of the hexosamine biosynthetic pathway, was overexpressed >2 fold in metastases compared to primary tumor (p = 0.007). Kaplan-Meier estimates from cBioPortal show that HCC patients with genetic alterations in GFPT2 and/or its isoform GFPT1 have significantly worse overall median survival compared to patients without these alterations (27.5 months vs 83.24 months, p = 0.006). Functional validation studies demonstrated that increased O-GlcNAcylation by pharmacologic treatment with TMG or genetic manipulation by GFPT2 overexpression was sufficient to increase migration and invasion in vitro in a panel of human and murine HCC cell lines. Decreased HBP-O-GlcNAcylation by pharmacologic treatment with DON or genetic manipulation by GFPT2 shRNA knockdown inhibited migration and invasion in vitro. Similarly, using a novel organoid invasion assay, TMG treatment increased invasion of organoids derived from LM tumors ex vivo 2-fold (p<0.03) while DON treatment inhibited invasion 2-fold (p<0.03). The hexosamine biosynthetic pathway is required and sufficient for invasion and metastasis in vitro in HCC cell lines and ex vivo in HCC tumor-derived organoids. The HBP may be a potential therapeutic target for advanced HCC patients.

Introduction: This study develops an innovative computational framework, Expression Variation Analysis (EVA), to model transcriptional dysregulation in cancer. Heterogeneity poses a major challenge in translational research. For example, inter-tumor heterogeneity limits the biomarker discovery and intra-tumor heterogeneity enables therapeutic resistance. Moreover, in some cancers driver mutations are insufficient to account for the widespread transcriptional variation responsible for these outcomes. Thus, new computational tools to model transcriptional variation are essential. Methods: EVA is a unified computational framework to model transcriptional variation in cancer. Briefly, EVA quantifies transcriptional heterogeneity for one set of samples or cells from one phenotype using the expected dissimilarity between pairs of expression profiles. U-statistics theory can then quantify the statistical significance of the difference in transcriptional heterogeneity between phenotypes. Results: We apply EVA to perform a comprehensive characterization of transcriptional variation in head and neck squamous cell carcinoma (HNSCC). At a pathway level, transcriptional variation in HNSCC tumors is higher than normal controls. Applying EVA to integrate ChIP-seq data with RNA-seq reveals that these pervasive transcriptional differences occur in enhancers. Similarly, applying EVA at a gene level to model splicing reveals more heterogeneity in transcript usage in tumor samples than normals. HPV- HNSCC tumors are unique in having mutations in genes that regulate the splicing machinery, and the HPV- tumors with these alterations have a greater number of dysregulated splice variants than those without. Nonetheless, the EVA analysis identifies a similar number of alternative splice variants in HPV+ as HPV- tumors suggesting an alternative mechanism of transcriptional heterogeneity in HPV+ disease. Adapting EVA to single cell data demonstrates that increased fibroblast composition is associated with greater variation in immune pathway activity in HNSCC. Moreover, we observe greater transcriptional heterogeneity in HNSCC primary tumors than lymph node metastasis consistent with a clonal outgrowth. Conclusions: We demonstrate that the statistical framework from EVA enables differential heterogeneity analysis in HNSCC ranging from pathway dysregulation, splice variation, epigenetic regulation, and single cell analysis. This algorithm provides a critical framework to model the hidden multi-molecular mechanisms underlying the complex patient outcomes that are pervasive in cancer. Citation Format: Bahman Afsari, Leslie Cope, Daria A. Gaykalova, Donald Geman, Sidharth Puram, Loyal A. Goff, Alexander Favorov, Elana Judith Fertig. Uncovering hidden sources of transcriptional dysregulation arising from inter- and intra-tumor heterogeneity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3399.

Abstract Despite the incorporation of the epidermal growth factor receptor (EGFR) inhibitor cetuximab into the clinical management of head and neck squamous cell carcinoma (HNSCC), limited to no long-term changes in overall survival are observed in HNSCC patients even though EGFR is expressed at high levels in these tumors. Therefore, the identification of novel therapeutic approaches to enhance the clinical efficacy of cetuximab could lead to improved long-term survival for HNSCC patients. Our previous work suggests that cetuximab activates the interleukin-1 (IL-1) pathway via tumor release of IL-1 alpha (IL-1α), although the implications of activating this pathway are unclear. The IL-1 pathway plays a central role in immune response and displays both pro-tumor and anti-tumor activities. IL-1 may promote tumor growth by upregulating the secretion of pro-inflammatory mediators involved in angiogenesis and metastasis. On the other hand, IL-1 signaling may promote antitumor immunity specifically via natural killer (NK)-cell mediated antibody-dependent cell-mediated cytotoxicity (ADCC), which is also an important mechanism of action of cetuximab. The goal of our work is to determine how modulation of the IL-1 pathway affects HNSCC tumor response to cetuximab. We found that blockade of IL-1 signaling using an IL-1-receptor antagonist (IL-1RA, anakinra), neutralizing IL-1α/IL-1β antibodies, and genetic knockdown of the IL-1R all suppressed the anti-tumor efficacy of cetuximab, while IL-1α overexpression and treatment with recombinant IL-1α enhanced HNSCC tumor response to cetuximab in immunodeficient and immunocompetent HNSCC mouse models. Mechanistically, these results appear to be due to modulation of ADCC, as we found that IL-1 blockade significantly reduced cetuximab-mediated ADCC in vitro. Additionally, we found that HNSCC patients with high baseline circulating levels of IL-1 ligands (IL-1α, IL-1β) were significantly more likely to respond favorably to cetuximab monotherapy compared to patients with low or no baseline circulating IL-1 ligands. Altogether, these results suggest that IL-1 signaling is necessary for HNSCC tumor response to cetuximab. Therefore, IL-1α warrants further study as a novel therapeutic to enhance response to cetuximab and as an immunologic biomarker for cetuximab response. Citation Format: Madelyn M. Espinosa-Cotton, Rachel A. Dahl, Elana Fertig, Isaac Jensen, Ayana J. McLaren, Kenley Miller, Samuel N. Rodman, Sandra Schmitz, Andrean L. Simons. Interleukin-1 signaling is required for HNSCC tumor response to cetuximab [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5734.

Alternative splice events (ASES) are significant components of potential oncogenic pathways alterations and play a critical role in malignant cell transformation in a variety of solid and liquid tumors, including head and neck squamous cell carcinoma (HNSCC). However, high throughput analyses performed to date have not considered ASEs. Therefore, they have detected a limited number of genetic alterations for HNSCC, which incompletely describe the HNSCC specific pathway alterations. The heterogeneous nature of these alterations has made the discovery of reliable HNSCC biomarkers and therapeutic targets for this disease challenging. We performed alternative splice events (ASEs) analysis to enhance our understanding of HNSCC biology. To define ASEs specific to HNSCC we designed a novel bioinformatics pipeline from RNA-sequencing data of HNSCC tumors and independent normal samples. Evaluating the top scoring candidates, we have found several highly promising ASE candidates, including GSN, Gelsolin, an actin-binding protein, a key regulator of actin filament assembly and disassembly. Previously published literature proposes that GSN demonstrates tumor-suppressor properties by reducing cell proliferation in vivo and in vitro via suppression of protein kinase C (PKC, part of the PI3K pathway which is altered in HNSCC). The alternative splicing event involves an insertion of 110 bp from the 14th intron. This insertion contains a stop codon in frame, and the splice variant gives rise to a truncated (562 amino acids(aa)) protein with only 4 Gelsolin domains (instead of the full-length 731 aa protein with 6 Gelsolin domains). Using RNA-Seq data we demonstrated that 40% of tumor samples harbor the GSN-ASE. QRT-PCR confirmed that while total expression of GSN is decreased in HNSCC samples, GSN is expressed in the alternative truncated form only in HNSCC tumors and not normal tissues. Accordingly, total GSN expression is also seen to be downregulated in breast, lung and colon cancers. Evaluation of TCGA data confirmed the pre-dominant expression of the truncated GSN isoform over the wt GSN in HNSCC, bladder urothelial carcinoma, colon adenocarcinoma, lung SCC, breast invasive carcinoma, cervical SCC and endocervical adenocarcinoma. Moreover, we confirmed that that the expression of the truncated GSN is important for the migration and invasion of the cancer cells in vitro. These data suggest that alternative splicing plays an important role in the GSN gene for multiple tumor types. Citation Format: Daria A. Gaykalova, Dylan Kelley, Theresa Guo, Craig Bohrson, Ilse Tiscareno, Veronika Zizkova, Michael Considine, Ludmila Danilova, Emily Flam, Justin Bishop, Julie Ahn, Samantha Merritt, Marla Goldsmith, Chi Zhang, Wayne Koch, William Westra, Zubair Khan, Michael Ochs, Sarah Wheelan, Elana Fertig, Joseph Califano. The discovery of novel GSN alternative splicing in head and neck squamous cell carcinoma. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2880.

A 30-UTR KRAS-variant is associated with cisplatin resistance in patients with recurrent and/or metastatic head and neck squamous cell carcinoma C. H. Chung1,2*, J. W. Lee3, R. J. Slebos4, J. D. Howard1, J. Perez1, H. Kang1, E. J. Fertig1, M. Considine1, J. Gilbert5, B. A. Murphy5, S. Nallur6, T. Paranjape6, R. C. Jordan9, J. Garcia10, B. Burtness7, A. A. Forastiere1 & J. B. Weidhaas6,8 Departments of Oncology; Otolaryngology-Head and Neck Surgery, Sidney Kimmel Cancer Center, Johns Hopkins University, Baltimore; Department of Biostatistics and Computational Biology, Dana-Farber Cancer Institute, Boston; Department of Cancer Biology; Division of Hematology/Oncology, Department of Medicine, Vanderbilt University, Nashville; Department of Therapeutic Radiology; Section of Medical Oncology, Department of Internal Medicine, Yale University School of Medicine, New Haven; Department of Radiation Oncology, University of California, Los Angeles; Department of Pathology, University of California, San Francisco; Mayo Clinic, Rochester, USA

A germline mutation in the 3'-untranslated region of KRAS (rs61764370, KRAS-variant: TG/GG) which disrupts microRNA regulation has previously been associated with altered patient outcome and drug sensitivity in various cancers. Our study suggests this KRAS-variant is a potential predictive biomarker for poor platinum response in recurrent/metastatic head and neck squamous cell carcinoma patients.A germline mutation in the 3'-untranslated region of KRAS (rs61764370, KRAS-variant: TG/GG) has previously been associated with altered patient outcome and drug resistance/sensitivity in various cancers. We examined the prognostic and predictive significance of this variant in recurrent/metastatic (R/M) head and neck squamous cell carcinoma (HNSCC).We conducted a retrospective study of 103 HNSCCs collected from three completed clinical trials. KRAS-variant genotyping was conducted for these samples and 8 HNSCC cell lines. p16 expression was determined in a subset of 26 oropharynx tumors by immunohistochemistry. Microarray analysis was also utilized to elucidate differentially expressed genes between KRAS-variant and non-variant tumors. Drug sensitivity in cell lines was evaluated to confirm clinical findings.KRAS-variant status was determined in 95/103 (92%) of the HNSCC tumor samples and the allelic frequency of TG/GG was 32% (30/95). Three of the HNSCC cell lines (3/8) studied had the KRAS-variant. No association between KRAS-variant status and p16 expression was observed in the oropharynx subset (Fisher's exact test, P = 1.0). With respect to patient outcome, patients with the KRAS-variant had poor progression-free survival when treated with cisplatin (log-rank P = 0.002). Conversely, KRAS-variant patients appeared to experience some improvement in disease control when cetuximab was added to their platinum-based regimen (log-rank P = 0.04).The TG/GG rs61764370 KRAS-variant is a potential predictive biomarker for poor platinum response in R/M HNSCC patients.NCT00503997, NCT00425750, NCT00003809.