Immunotherapy Resistance Research Articles

CDKN2Alow tumours mediate immunotherapy resistance by depriving macrophages of zinc.

CDKN2Alow tumours mediate immunotherapy resistance by depriving macrophages of zinc.

Abstract 3279: Smoking burden is associated with immunotherapy resistance in extensive stage small cell lung cancer

Abstract Small cell lung cancer (SCLC) remains the deadliest histologic lung cancer subtype with a median survival of only ∼12 months among patients with extensive stage (ES) disease. SCLC is strongly associated with heavy tobacco use, yet the effect of tobacco use on therapeutic outcomes remains understudied. Recent data from our group showed that the subtype of SCLC with inflammatory gene signatures (SCLC-I) responds better to immunotherapy (IO) than the other subtypes. Given the association between smoking and inflammation, we hypothesized that heavy smoking may correlate with better IO responses and enrichment in the SCLC-I subtype. To investigate this, we evaluate the outcomes of ES-SCLC patients treated at MD Anderson between 2017-2024 (N=521) and compared those who received chemotherapy plus IO (N=263) to those who received chemo alone (N=258). Transcriptional changes were evaluated across different tobacco pack-year cohorts using the GEMINI bulk RNAseq database of MD Anderson SCLC tumors (N=81). The median overall survival (mOS) for all ES-SCLC patients was 13.3 months. Immunotherapy provided an added 1-month benefit in mOS over chemotherapy alone, extending mOS from 12.1 to 13.1-months (HR=0.88; p=0.21); consistent with historic cohorts. Patients were then stratified by the cohort pack-year median of 40. For patients with ≤ 40 pack-year smoking histories, the addition of IO improved mOS by 3.7-months from 11.5 to 14.3-months (HR=0.74; p=0.03). However, those with >40 pack-year smoking histories demonstrated no benefit from added IO (13.0 to 12.2-months (HR=1.06; p=0.70). Further analysis by pack-year tertiles (0-24, 25-49, 50+) showed mOS benefits of 5.4 months (HR: 0.67, P=0.05), 1.6m (HR:0.91, P=0.62), and 0.7m (HR=1.01, P=0.679) respectively, for chemo plus IO vs chemo alone. Despite better IO outcomes among patients with 0-40 pack-years, the SCLC-I subtype was underrepresented in the 0-40 pack-year group compared to expectation using the GEMINI RNAseq database (chi-squared 8.2; p=0.04). Tobacco exposure positively correlated with absolute immune cell infiltration by CIBERSORTx (R=0.24; p=0.03), although driven by higher immunosuppressive T cell signatures. Tumors from the 50+ pack-year (upper smoking tertile) group also demonstrated elevated chronic inflammatory GSEA signatures and genes associated with microenvironment immune suppression compared to the 0-24 (lower smoking tertile) pack-year group. The discovery that smoking is linked to immunotherapy resistance in SCLC represents a novel contribution to our understanding of the disease and its treatment. By identifying immunosuppressive cellular infiltrates and immunosuppressive gene signatures among patients with heavy vs light smoking histories, we have identified potentially targetable mechanisms of immune evasion. This represents a new lens through which to study immunotherapy resistance in SCLC. Citation Format: Kyle Concannon, Benjamin Morris, Lixia Diao, Moushumi Sahu, Whitney Lewis, Xiaowen Sun, Simon Heeke, Bingnan Zhang, Hai Tran, Lingzhi Hong, Waree Rinsurongkawong, Minh Nguyen, Kang Qin, Hui Li, Jainjun Zhang, Jack Lee, Jing Wang, Robert Cardnell, Carl Gay, John Heymach, Lauren Byers. Smoking burden is associated with immunotherapy resistance in extensive stage small cell lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3279.

Abstract 635: Unraveling the potential of neogenin 1 in non-small cell lung cancer: A path toward innovative combination therapies

Non-small cell lung cancer (NSCLC) has a poor survival rate due to delayed diagnosis and treatment resistance. Currently, immunotherapy has advanced the NSCLC treatment; however, only a small fraction of patients respond to the therapy, demanding the identification of novel biomarkers associated with resistance to develop new therapeutic strategies. Herein, we investigated the role of neogenin 1, a transmembrane protein that belongs to the immunoglobulin superfamily, in seven different human NSCLC cell lines and compared the expression with Jurkat T cell line and human peripheral blood mononuclear cell (PBMC)-derived CD3/CD28-activated CD8+ T cells using qPCR and western blot. We also performed the cell proliferation assay in adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) human cell lines after knockdown and overexpression of neogenin 1. We found that neogenin 1 was highly expressed in primary CD8+ T cells compared to the cancer cell lines, implying that genetic alteration of neogenin 1 might interfere more with CD8+ T cell effector function. We also discovered that both adenocarcinoma and squamous cell carcinoma cell proliferation ability were influenced by genetic alteration of neogenin 1. In addition, gene ontology enrichment analysis and KEGG pathway analysis indicated that neogenin 1 was involved in the TGF-beta signaling pathway, possibly suggesting that neogenin 1 might play a role in therapeutic resistance, including immunotherapy, in NSCLC. Correlation analysis in LUAD and LUSC in the TCGA dataset also revealed that neogenin 1 was positively correlated with a set of genes, such as IGF1R and USP8, associated with immunotherapy resistance. Further investigation of this potential gene in vivo will unfold its role in immunotherapy resistance in NSCLC, paving the way for developing novel and effective combinatorial therapeutic strategies targeting neogenin 1. Citation Format: Asmita Pandey, Shaoqiu Chen, Li Ma, Mayumi Jijiwa, Masaki Nasu, Vedbar Khadka, Yu Chen, Hua Yang, Sudhir Rai, Youping Deng. Unraveling the potential of neogenin 1 in non-small cell lung cancer: A path toward innovative combination therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 635.

Abstract 113: Decoding developmental programs driving tumorigenesis and immunotherapy resistance in triple-negative breast cancer

Abstract Background: Triple-negative breast cancer (TNBC), defined by a lack of hormone receptors and lack of HER2 amplification, is the deadliest subtype of breast cancer. TNBC kills 1 in 3 patients and while promising new drugs based on PARP inhibition and anti-PD1 immunotherapy can extend survival in selected patients, 30-40% of patients relapse or fail therapy. Increasing evidence suggests that TNBC tumors harbor stem-like tumorigenic cancer cells with high plasticity, capable of replenishing cancer cell populations and linked to adverse outcomes. Unfortunately, the phenotypic diversity, microenvironmental contexture, and functional significance of these cells remain poorly understood and available marker genes are not specific, precluding rational drug development. Methods: We created a multimodal single-cell RNA-seq atlas of 80 TNBC patients, including clinical covariates such as BRCA1/2 mutations, race/ethnicity, and prior treatment status. We then defined differentiation-associated malignant cell states using a novel interpretable deep learning approach for determining single-cell potency. We subsequently interrogated geospatial features, including localized microenvironments of less differentiated malignant cells, and linked them to immune checkpoint inhibitor (ICI) response. Results: Our quantitative analysis shows that TNBC is enriched in cancer cells with high developmental potency as compared to other breast cancer subtypes, consistent with its aggressive phenotype. Critically, we observed a strong association between enrichment of high potency cells and resistance to ICI treatment across human and mouse bulk expression data, both within TNBC and across other cancer types, a finding that we experimentally validated in a murine model of TNBC. In scRNA-seq and spatial transcriptomics data, we defined striking relationships between less differentiated cancer cells and context-dependent T cell states indicative of strong immunosuppressive capability. In line with this, we found that knockout of genes associated with immature cancer cells promotes TNBC cell killing by T cells in a co-culture CRISPR screen. Ongoing experiments include perturbations of novel therapeutic targets emerging from this work, with the goal of improving immunotherapy responses and ultimately extending patient survival. Conclusion: We anticipate that our approach will expand our understanding of developmental landscapes in TNBC and lead to new opportunities for developing targeted drugs in this devastating disease. Citation Format: Rachel Gleyzer,Farnaz Khameneh,Zhen Qi,Wubing Zhang,Chloé B. Steen,Minji Kang,Maya Maalouf,Melanis Ghadimi,Makenna Lindsey,Sally Bobo,Thomas J. Lomis,Frederick M. Dirbas,Michael F. Clarke,Aaron M. Newman. Decoding developmental programs driving tumorigenesis and immunotherapy resistance in triple-negative breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 113.

Abstract 916: Tumor extracellular vesicles trigger a dendritic cell stress response to promote immune evasion and immunotherapy resistance

Abstract Previous studies have demonstrated dendritic cell (DC)-mediated T cell activation to be impaired within the tumor microenvironment (TME). This has included studies showing that activation of XBP1 within the unfolded protein response (UPR) pathway triggers lipid accumulation by DCs within the TME, however the signal mediating XBP1 activation has remained unclear. We have previously demonstrated that fatty acid oxidation (FAO) as well as the activation of the SREBP2-dependent mevalonate biosynthetic pathway leads to the development of mregDCs in the TME, exhibiting impaired antigen cross-presentation and an enhanced capacity to drive regulatory T cell (Treg) differentiation. Despite the critical role of the DC in orchestrating anti-tumor immunity, there are currently no therapeutic strategies to augment their function. Using an autochthonous model of BRAFV600E melanoma, we demonstrate that tumor-dependent generation of extra-cellular vesicles (ECVs) suppresses anti-tumor immunity and supports tumor progression. Using bulk and single-cell RNA sequencing analysis, we show that tumor ECVs upregulate various components of the UPR pathway in conventional DCs (cDCs), including IRE1alpha-XBP1 and SREBP2 activation based on qPCR and Western blot studies. We further show that XBP1 activation leads to PPARalpha nuclear translocation and enhanced FAO. We verify that lymph node DC populations engulfing labeled tumor ECVs harboring a CD81-GFP fusion protein also exhibit both enhanced XBP1/PPARalpha activation as well as the transcriptional signature of mregDCs. Together, these ECV-dependent effects suppress DC-mediated antigen cross-presentation and CD8+ T cell activation while promoting DC-dependent Treg differentiation in vitro and in vivo. Additional studies were conducted showing similar effects generated by ECVs derived from models of colon and breast cancer. We subsequently found DC-specific knock-out of PPARalpha to partially reverse the effects of tumor ECVs on DC functionality in an engineered transgenic murine model. Indeed, a PPARalpha antagonist, TPST-1120, overcomes resistance to anti-PD-1 immunotherapy in a poorly immunogenic autochthonous melanoma model while suppressing DC lipid stores and DC FAO in situ. Together, this work suggests that pharmacologic targeting of pathways elicited by tumor-derived ECVs can reverse DC tolerization in the TME and overcome immunotherapy resistance in select tumors. Understanding which tumors rely on ECVs to suppress anti-tumor immunity will be a critical step in prospectively developing targeted therapeutics to block these immune evasive pathways and enhance the efficacy of checkpoint inhibitor immunotherapy. Citation Format: Mandy Wang, Michael Plebanek, Kaylee Villarreal, Nicholas C. DeVito, Bala Theivanthiran, Brent A. Hanks. Tumor extracellular vesicles trigger a dendritic cell stress response to promote immune evasion and immunotherapy resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 916.

Abstract 6391: OMTX705: A dual-mechanism ADC targeting FAP-positive CAFs, to overcome chemo and immunotherapy resistance in solid tumors

Abstract ADCs targeting the tumor stroma offer a promising strategy to address tumor heterogeneity. Traditional tumor-targeting ADCs face limitations, as intratumor and interpatient variability in target protein expression can result in incomplete targeting. In contrast, the tumor stroma is more genetically stable and consistent across patients. CAFs, the predominant stromal cell type within the tumor microenvironment, play a critical role in cancer progression and immune evasion in multiple cancers. OMTX705, a clinical-stage ADC, targets FAP-positive CAFs in the tumor microenvironment. OMTX705 presents an optimized and highly stable linker and it is quickly internalized and transported to late endosomal compartments in FAP-expressing cells. Its novel cytolysin-based payload demonstrates strong cell killing activity upon specific ADC internalization in FAP positive cells. Indeed, enhanced cytotoxicity was observed when compared to MMAE, particularly in cells with high PgP expression, offering a potential solution for tumor drug resistance. OMTX705 has demonstrated preclinical robust efficacy, achieving 100% tumor growth inhibition and even regression, in PDX murine models for pancreatic, gastric, ovarian, triple negative breast and lung cancers, with or without humanized immune systems, and showing no significant toxicity. Higher anti-tumoral effect was evidenced over both classical chemotherapy and modern immunotherapy drugs, as a single agent or in combination therapies with Gemcitabine/Abraxane or immune checkpoint inhibitors. OMTX705 even displays superiority compared to other FAP-targeted ADCs bearing different payloads such as deruxtecan in pancreas cancer models. Moreover, it exhibits high plasma stability and excellent safety profile, with no adverse effects observed in non-human primates at doses up to 100 mg/kg. OMTX705 operates through a dual mechanism of action. First, upon ADC internalization, the bystander effect of the payload released from FAP-positive CAFs leads to efficient killing of neighboring tumor cells. This effect has been validated through different experimental techniques in vitro and in preclinical tumor samples. The second mechanism involves immune cell infiltration within the tumor microenvironment, further strengthening its therapeutic potential. This dual effect, combined with a favorable safety profile, positions OMTX705 as a promising candidate among cancer therapeutics, targeting the stroma in solid tumors and enhancing antitumor immune responses. Citation Format: Patricia González, Isabel Egaña, Cristina Ferrer, Roland Kontermann, Oliver Seifert, Klaus Pfizenmaier, Wolfgang Richter, Laureano Simon, Myriam Fabre. OMTX705: A dual-mechanism ADC targeting FAP-positive CAFs, to overcome chemo and immunotherapy resistance in solid tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6391.

Abstract 6466: Discovery Proteome and Whole Transcriptome Atlas analysis of skin cancer reveals biomarkers associated with immunotherapy response and resistance

Cutaneous squamous cell carcinoma (cSCC) is the second most common skin cancer and represents a major global health burden. Approximately 50% of cSCC patients develop primary resistance and 20% will develop secondary resistance to immune checkpoint inhibitors (ICI). There are limited biomarkers that reflect the tumor dynamics predictive of response to ICI therapy and we need to develop a better understanding of the cSCC tumor microenvironment (TME). Enabling simultaneous analyses of proteins and RNAs at ultra-high plex in spatial context, the GeoMx Digital Spatial Profiler (DSP) offers an advanced same slide multiomics workflow and is uniquely positioned to help unravel cellular dynamics throughout the cSCC TME. In this study, we spatially profiled over 1000 proteins (GeoMx Human Discovery Proteome Atlas (DPA)) and over 18,000 transcripts (GeoMx Whole Transcriptome Atlas) using the DSP on formalin fixed paraffin embedded sections of pre-treatment cutaneous skin cancers prior to immunotherapy. Pathology-informed, tumour and stromal regions were sampled across whole tissues to generate histologically relevant assessments of the skin tissues. Covering areas such as immuno-oncology, oncology, immunology, epigenetics, metabolism, cell death, and signaling pathway regulation, the DPA panel supported the discovery of biomarkers and potential therapeutic targets. Our study identified unique tumour and stromal signatures and biomarkers associated with both resistance and sensitivity to immunotherapy at both protein and RNA resolutions. Moreover, coverage across the protein hallmarks of cancer provided deeper insights into the tumour microenvironment dictating clinical responses to therapy. Our study provides an integrated pipeline for discovery based spatial proteomic and transcriptomic assessments of skin cancers. Citation Format: Arutha Kulasinghe, Christine Kang, Hiromi Sato, Brian Filanoski, Lori Hamanishi, Ashley Heck, Margaret Hoang, Alyssa Rosenbloom, Michael Rhodes, Joseph M. Beechem. Discovery Proteome and Whole Transcriptome Atlas analysis of skin cancer reveals biomarkers associated with immunotherapy response and resistance [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6466.

Abstract 911: Cancer stem cells establish intrinsic immune resistance through suppressing IFN-mediated antitumor immunity

Abstract Although the cancer immunotherapy has generated promising outcomes in the clinics, a substantial proportion of cancer patients either show resistance to immunotherapy or experience relapse following an initial response. This issue is particularly urgent for patients with squamous cell carcinomas (SCCs), highlighting the critical need to understand the mechanisms behind immune evasion and immunotherapy resistance in these cancers. Recently, our lab identified a subset of tumor-initiating stem cells (tSCs) in skin and head/neck SCCs that respond to transforming growth factor β (TGF-β) and play a key role in driving tumor relapse post-immunotherapy. Notably, these tSCs are resistant to T-cell attacks despite retaining functional antigen presentation, resulting in only transient responses to immune checkpoint blockade therapy across SCC types. Here, my current study has uncovered a central role for Sox2 as a master regulator that drives immune resistance in TGF-β-responsive tSCs. Through RNA-seq and qPCR analyses, we observed that Sox2 amplification in tSCs significantly blunted their interferon-γ (IFN-γ) responsive potentials. Mechanistically, we found that Sox2 neither blocks IFN-γ signaling transduction nor directly binds to the enhancers or promoters of most IFN-γ-induced genes. Instead, we discovered that the reduced IFN-γ response in Sox2-overexpressing cells is likely due to the altered Stat1-mediated transcriptional repression. This study uncovered a stem cell specific transcription regulation that provides new insights in developing new strategies to abolish tSC-specific immune resistance. Citation Format: Xuejie Huang, Weijie Guo, Yuxuan Phoenix Miao. Cancer stem cells establish intrinsic immune resistance through suppressing IFN-mediated antitumor immunity [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 911.

Abstract 3819: Modeling T cell and cancer cell interactions to examine immune checkpoint inhibition resistance in pancreatic cancer

Abstract Pancreatic cancer (PC) is a lethal cancer, with ∼90% of patients dying within 5 years. This dire statistic underscores the limitations of current treatments and the need for improved therapeutic approaches. Immune checkpoint inhibition (ICI) has shown promise in promoting long-term progression-free survival across various cancers by reactivating a patients’ latent anti-cancer immune response. However, PC is almost completely resistant to ICI, highlighting the incomplete understanding of immune-cancer interactions in this disease. The tumor microenvironment is a key factor in immunotherapy resistance in PC, where immunosuppressive immune cells and cancer-associated fibroblasts (CAFs) act to inhibit T cell function. Additionally, the extracellular matrix (ECM) landscape of PC presents a physical barrier to T cell infiltration and motility within the tumor. Developing strategies to overcome these barriers has been challenging, partly due to the difficulty in replicating the complex PC tumor microenvironment in experimental models. The use of 3D cancer tumoroid cultures enables modeling of cancer and stromal cells spatial arrangements, mimicking native tumor microenvironments and facilitating high resolution examination of cell-to-cell and cell-to-environment interactions. However, most tumoroid cultures often use commercial matrix products like basement membrane extracts (BME), which poorly recapitulate the distinct biochemical and mechanical properties inherent to PC tumors. Here, we utilize the Inventia RASTRUMTM platform to create 3D cancer/CAF/T cell co-cultures based on synthetic PEG-based hydrogel matrices, which were tuned to mimic the stiffness and ECM composition of PC tumors. Using real-time longitudinal imaging and 3D image analysis pipelines, we visualize cancer-stromal cell interactions, which are obscured in conventional BME-based models. We also investigate the role of CAFs in supporting immune evasion and highlight the effect of multiple immunomodulators on cancer/T cell behavioral dynamics. Overall, our approach provides a scalable in vitro framework to dissect cell-to-cell interactions and offers a platform to the profile immunomodulatory properties of drugs. Future work using this platform will focus on molecular profiling of cancer-CAF-T cell interactions to provide predictive data aiding the rational design and evaluation of ICI combination therapies prior to pre-clinical in vivo work and clinical translation. Citation Format: Aji Istadi, Inna Navarro, Silvia Lombardi, Dannielle Upton, Mezzalina Vankan, Cesar Moreno, Michael Trpceski, Cecilia Chambers, David Hermann, Marco Herold, Paul Timpson, Sean Porazinski, Greg Neely, Marina Pajic. Modeling T cell and cancer cell interactions to examine immune checkpoint inhibition resistance in pancreatic cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3819.

Abstract 5296: Hepatocellular carcinoma resistance and recurrence to anti-PD-1 therapy is associated with tumor heterogeneity and plasticity driven by MYC

Abstract Hepatocellular carcinoma (HCC) is the most prevalent type of liver cancer and a major cause of cancer-related deaths worldwide. HCC is characterized by low survival rates and a high incidence of recurrence. Systemic therapy responses are observed in only a small proportion of HCC patients. Our spatial transcriptomics (ST) analysis of HCC tumors from patients treated with anti-PD-1 and cabozantinib (a multi-tyrosine kinase inhibitor) suggested that immunotherapy resistance is driven by cancer cell heterogeneity, tumor microenvironment (TME) dynamics, and cancer cell plasticity. These mechanisms were detected in an HCC sample collected from a responder who later experienced recurrence. To understand the genomic drivers of resistance and recurrence in HCC, we performed new bioinformatic analyses to investigate the molecular mechanisms associated with HCC plasticity that result in anti-PD-1 resistance. Across our ST samples, we observed up-regulation of oncogenic and immune evasion mechanisms following anti-PD-1 treatment. Interestingly, these signaling pathways share a common upstream modulation by MYC. Gene set enrichment analysis identified a significant upregulation of the MYC pathway in immune-poor cancer regions. Spatially weighted correlation analysis revealed that MYC upregulation is colocalized with cancer cell-intrinsic pathways, such as proliferation, metabolism, and cell plasticity, and is spatially exclusive with immune-related pathways, including antigen presentation and interferon-gamma response. Moreover, intercellular interaction analysis showed that MYC-upregulated cancer cells engage in crosstalk with cancer-associated fibroblasts (CAFs) via PDGF-PDGFR signaling, which is associated with the epithelial-mesenchymal transition (EMT) process, a hallmark of cell plasticity. Our findings were validated in additional anti-PD-1-treated HCC single-cell and ST public datasets, confirming that therapy-reinvigorated lymphocytes neutralize only MYC-downregulated HCC cells, leaving MYC-high HCC cells to survive. These residual HCC cells potentially become the source of tumor recurrence. Our findings highlight MYC as a key regulator of therapeutic resistance and immune suppression in HCC, potentially serving as a target for future interventions to delay anti-PD-1 treatment relapse and improve patient outcomes. Citation Format: Shuming Zhang, Ludmila Danilova, Jacob T. Mitchell, Iris Kwon, Aleksander S. Popel, Mark Yarchoan, Elizabeth M. Jaffee, Elana J. Fertig, Luciane T. Kagohara. Hepatocellular carcinoma resistance and recurrence to anti-PD-1 therapy is associated with tumor heterogeneity and plasticity driven by MYC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5296.

Abstract 4250: Tumor-wide RNA splicing aberrations generate immunogenic public neoantigens across cancers

Abstract Background: High tumor heterogeneity and low mutational burden in cancers pose significant challenges for immunotherapy. To address this, we developed a novel in silico pipeline to characterize cancer-specific splicing events (neojunctions; NJs) expressed ubiquitously across ten cancer types. This approach successfully identified tumor-wide, public, alternatively splicing neoantigens (ASNs) that elicit CD8+ T-cell-mediated cytotoxicity across multiple cancers. Methods: Our SSNIP pipeline identified recurring public NJs expressed in TCGA RNA-seq data (positive sample rate (PSR) > 10%) but not in GTEx normal tissue RNA-seq data (PSR < 1%) across 10 cancer types. We utilized multi-site RNA sequencing data to characterize intratumorally conserved NJs. Our in-house glioma dataset included 56 patients with approximately 10 spatially distinct intratumoral biopsy sites per patient (n=535). In vitro sensitization (IVS) of CD8+ T-cells from healthy donors against high-confidence ASN candidates, followed by 10x VDJ scRNA-seq, identified ASN-specific TCR sequences. TCR-transduced donor-derived CD8+ T-cells were cultured against NJ-expressing glioma and melanoma cell lines to evaluate tumor-specific killing. CRISPRi KD of disease-specific splicing factors was performed to assess NJ modulation. Results: Our pipeline identified 789 public NJs, with 32 NJs concurrently identified in transcriptomic and proteomic data and predicted to be presented by HLA-A*02:01 with high confidence. We captured TCR clonotypes reactive against NJs in RPL22 (n=7) and GNAS (n=1), the latter being highly intratumorally-conserved (detected in > 90% of spatially-mapped biopsies across 17/56 patients (26.78%)). Mutant GNAS-derived ASN-specific CD8+ T-cell clones were detectable from glioma patient PBMC by IVS. TCR-transduced T-cells demonstrated recognition and tumor-specific killing against endogenously processed and presented ASNs in multiple glioblastoma and melanoma cell lines. Furthermore, IDH1-mutant oligodendroglioma samples demonstrated significantly elevated expression of NJs over IDH1-mutant astrocytoma and IDH1wt subtypes. Differential gene expression identified decreased expression of splicing factors due to oligodendroglioma-specific co-deletion of Chromosomes 1p/19q. CRISPRi KD of these splicing factors (e.g. SF3A3, SNRPD2) in IDH1wt glioma cells resulted in significantly increased expression of corresponding NJs. Performing DESeq2 on the remaining cancer types revealed similar splicing gene set-associated differences. Conclusions: SSNIP identified novel public tumor-wide splice-derived neoantigen candidates and ASN-specific TCRs, offering a promising off-the-shelf immunotherapy approach for diverse cancer types. Characterization of intratumorally conserved neoantigens addresses the critical challenge of intratumoral heterogeneity in immunotherapy resistance. Citation Format: Darwin Kwok, Nicholas Stevers, Inaki Etxeberria, Takahide Nejo, Maggie Colton Cove, Lee Chen, Jangham Jung, Kaori Okada, Senthilnath Lakshmanachetty, Marco Gallus, Abhilash Barpanda, Chibo Hong, Gary Chan, Jerry Liu, Samuel Wu, Emilio Ramos, Akane Yamamichi, Payal Watchmaker, Hirokazu Ogino, Atsuro Saijo, Aidan Du, Nadia Grishanina, James Woo, Aaron Diaz, Susan Chang, Joanna Phillips, Arun Wiita, Christopher Klebanoff, Joseph Costello, Hideho Okada. Tumor-wide RNA splicing aberrations generate immunogenic public neoantigens across cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr XX.

Abstract 4536: Dissecting tumor-immune microenvironment in response and resistance to immune checkpoint blockade in metastatic melanoma

Abstract Background: Immune checkpoint inhibitor (ICI) therapies have markedly improved the prognosis for patients with stage III & IV metastatic melanoma by prolonging progression-free and overall survival rates. However, the variability in immune evasion and resistance mechanisms presents significant challenges to the clinical efficacy of ICIs. This project aims to define drivers of immunotherapy response and resistance by employing advanced genomic, single-cell mRNA analyses, and spatial profiling techniques on tissue biopsies from metastatic melanoma patients. Methods: In this study, we developed a framework to analyze response and resistance, both intrinsic and acquired, via immune features in the tumor microenvironment in a standardized, uniformly processed, and deeply clinically annotated cohort of metastatic melanoma patients (n=61) treated with ICB as part of the human tumor atlas network (HTAN) initiative. From the tumor samples, we conducted single-nucleus RNA sequencing, and for a subset of the samples high-resolution spatial imaging (including protein mIHC, CODEX, and transcriptomics MERFISH). Standardized processing and data pipelines allowed for integrating genomic, transcriptomic, and spatial features to elucidate characteristics and mechanisms in tumor microenvironment and their relationships with resistance. Results: Studies of the pretreatment samples demonstrated that CD4 and CD8 T cells, particularly TCF7+ CD8 T cells, are significantly more prevalent in responders to immunotherapy. Conversely, macrophages, especially Angio-TAMs, show higher levels of enrichment in non-responders. Moreover, the presence of B cells and follicular dendritic cells in non-lymph node biopsies supports the presence of tertiary lymphoid structures within the TME. Three levels of immune enrichment were identified through spatial analysis, and samples with more immune enrichment tend to have better responses. We also identified 10 recurrent cellular neighborhoods (RCNs) and found that RCN2,4, and 7 with high lymphocyte infiltration are significantly more enriched in responders than non-responders. In addition, RCN4 is associated with immune infiltration while RCN7 is TLS-like. Conclusions: Our findings indicate that patients with low immune infiltration exhibited enrichment of macrophages associated with the hypoxia and angiogenesis phenotypes, while patients with high immune infiltrates displayed enrichment in lymphocytes, particularly TCF7+ CD8+ T cells, confirming previous findings and indicating a robust T cell-mediated immune response. This project integrates genomic, transcriptomic, and spatial features to elucidate shared tumor and microenvironmental states and their relationships with resistance, and guide more personalized and effective treatment strategies for metastatic melanoma. Citation Format: Xinyu Cui, Giuseppe Tarantino, Yiwen He, Priyanka Solanky, Kathleen Pfaff, Aaron R. Thorner, Tyler J. Aprati, Boyang Zhang, Timothy Blosser, Emily Robitschek, Jiajia Chen, Junko Tsuji, Elliot Boblitt, Allison Frangieh, Hannah M. Faulkner, Marta Holovatska, Aleigha Lawless, Michael Manos, Karla Helvie, Tatyana Sharova, Dennie Frederick, James L. Fahey, Diego Villamarin, Sachi Krishna, Chanell Mangum, Ajit J. Nirmal, Domenic Abbondanza, Cai McCann, Bruce Johnson, Alex K. Shalek, Eliezer Van Allen, Xiaowei Zhuang, Ryan Sullivan, Barbara E. Engelhardt, Samouil L. Farhi, Scott Rodig, F. Stephen Hodi, Genevieve M. Boland, David Liu. Dissecting tumor-immune microenvironment in response and resistance to immune checkpoint blockade in metastatic melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4536.

Abstract 179: Spatial and multiomics analysis reveals immune interactions as key drivers of immunotherapy outcomes in melanoma patients with in-transit metastases

Abstract Background: In-transit metastasis (ITM) of melanoma occurs between the primary tumor and draining lymph nodes. ITM can be managed by by adjuvant immunotherapy after resection, or immunotherapy alone when unresectable. Nevertheless, nearly half of ITM patients are resistant, which constitutes a clinical challenge. Multiomic spatial profiling offers insights into features influencing immunotherapy response, but the tumor microenvironment (TME) and biological determinants of ITM remain poorly defined. This study aims to characterize molecular and spatial TME attributes in ITM patients to improve understanding of immunotherapy response and resistance. Methods: Fifty-four samples from ITM patients treated with immunotherapy were collected at pre-treatment (PRE) and progression (PROG). Patients with a complete or partial response in the advanced setting, or recurrence-free survival over 12 months in the adjuvant setting, were classified as "responsive," and others as "resistant". Single-cell sequencing and 40-marker imaging were conducted on 18 matched tumor dissociates and FFPE samples at PRE (n=3 resistant, n=5 responsive) and PROG (n=10). Whole genome sequencing on 12 fresh frozen samples (n=4 resistant, n=8 responsive at PRE) were used to identify somatic variants. Results: While regional immune cell proportions were broadly similar across response groups, PRE responders showed increased intra- and peri-tumoral T cells, and surprisingly, reduced CD141+ dendritic cells (cDC1s) compared to resistant patients (P=0.024). K-means clustering identified 7 spatially resolved cellular neighborhoods (CNs) across all ITMs with heterogeneous cellular constitution; nevertheless, lymphocyte-enriched CNs were more dispersed in PRE responders and interacted with other immune and tumor CNs, while remaining peripheral in PRE resistant tumors. Despite having a lower DC density, pairwise interaction analysis highlighted significant interactions between cDC1s and CD14+ macrophages in PRE responders, and between macrophages and IDO1+ CD4+ T cells or Granzyme B+ CD8+ T cells (P<0.05, log2 fold change>0.2), suggesting enhanced macrophage-driven local T cell stimulation in the absence of DCs. Single-cell transcriptomics analysis comfirmed the reduced proportion of cDC1s in PRE responders compared to resistant ITM (P=0.006). Genomic profiling revealed EREG mutations exclusively in resistant ITMs, which has been linked to immunoregulationand tumor invasion through the EGFR pathway. Ongoing analysis explores interactions among DCs, macrophages, T cells, B cells, melanocytes, and endothelial cells to elucidate TME dynamics associated with immunotherapy outcomes. Conclusion: These findings deepen our understanding of tumor-immune interactions in ITM patients and inform personalized immunotherapy approaches tailored to their unique TME. Citation Format: Xinyu Bai, Camelia Quek, Ghamdan Al-Eryani, Sonny Ramkomuth, Louise Baldwin, Jiabao Tian, Nurudeen A. Adegoke, Ismael A. Vergara, Felicity Newell, Kate Harvey, John Reeves, Andrew J. Spillane, Nicola Waddell, Serigne N. Lo, Alexander M. Menzies, Ines P. da Silva, Umaimainthan Palendira, Georgina V. Long, Richard A. Scolyer, Alexander Swarbrick, James S. Wilmott. Spatial and multiomics analysis reveals immune interactions as key drivers of immunotherapy outcomes in melanoma patients with in-transit metastases [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 179.

Abstract 7297: Enhanced efficacy of CTX-471, a CD137 agonist antibody, in models of immune checkpoint failure via simultaneous blockade of neo-angiogenesis

Abstract CTX-471 is a monoclonal agonistic anti-CD137 antibody currently under investigation in solid tumors. To increase the breadth and duration of response of CTX-471 we set out to investigate its anti-tumor efficacy in combination with anti-angiogenesis agents in murine models. CTX-009 is a bispecific antibody targeting VEGF-A and DLL4 which is undergoing clinical investigation in multiple cancer types, including biliary tract and colorectal cancers. Encouraged by early findings showing activity of the combination in a B2m negative mouse tumor model, engineered to mimic HLA loss in patients, we broadened the exploration of the combinatorial treatment in an additional B2m knockout model using a different murine genetic background (CT26B2m-/-, Balb/c). Furthermore, we developed two novel mouse models of immunotherapy resistance without the enhanced NK cell susceptibility bias conferred by complete MHC-I loss due to B2m deletion. CT26 cells with an engineered deletion of the B2m gene were passaged in CT26 tumor-experienced mice and a line of CT26B2m-/- cells that escaped immune rejection was established (CT26 B2m knockout escapers, CT26B2m-/-E). For the second model, the H-2k1 MHC-I locus was knocked out in the MC38 cells, resulting in targeted homozygous loss. Using this approach, the expression of the other MHC-I alleles, and therefore natural resistance to NK cells, are expected to be left intact. Results: Coupling VEGF-A/Dll4 targeting with CD137 agonism markedly increased the anti-tumor efficacy of the individual treatments alone in the MC38B2m-/-, MC38H-2k1-/-, CT26B2m-/- models as well as in the highly IO resistant CT26B2m-/-E model. Tumor growth inhibition was accompanied by pharmacodynamic evidence consistent with decrease of vascular bed density and enhancement of cellular cytotoxic immunity. CTX-471 monotherapy was still efficacious, albeit blunted, in these IO resistant models. The latter preclinical results mirrored data from a Ph1 clinical trial of CTX-471, where evidence of NK activation following CTX-471 treatment was seen in both blood and tissue samples of patients with known defects in the antigen presentation machinery. The findings presented here suggest that an agonistic anti-CD137 antibody may provide enhanced clinical benefit when employed alongside an anti-angiogenesis agent, such as CTX-009, in tumors where immune checkpoint inhibitors had previously failed. Citation Format: Amy Daniel Ulumben, Diana I. Albu, Rachael Duffy, Tristan Lubinski, Jason Kong, Jessica Zolotaresvsky, Melissa Brundin, Mei Su, Ruturaj Jadhav, Bing Gong, Thomas Schuetz, Kris Sachsenmeier, Alberto Visintin. Enhanced efficacy of CTX-471, a CD137 agonist antibody, in models of immune checkpoint failure via simultaneous blockade of neo-angiogenesis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7297.

Abstract 3852: The role of endothelial cell remodeling in driving immunotherapy resistance of hepatocellular carcinoma

Introduction: Hepatocellular carcinoma (HCC) is the most common liver cancer and a leading cause of cancer-related deaths globally. Immune checkpoint blockade (ICB) therapy has shown promise but has yielded limited responses in HCC patients. Combining anti-vascular endothelial growth factor (VEGF) with ICB enhances therapeutic efficacy, however, a deeper understanding of endothelial cell (EC) remodeling is crucial for further improving ICB effectiveness. Our integrated analyses identified a transcription factor-4 (TCF4)-driven transcriptional program, coupled with elevated BRD4 expression, associated with EC remodeling in ICB-resistant HCC. Emerging studies have shown that TCF4 functions depend on BRD4 through recruitment to a specific enhancer region, and BRD4 inhibitors can reduce TCF4 expression. We aim to delineate the microenvironmental cue and transcriptional network that direct EC remodeling for tumor immunosuppression and their role in promoting ICB resistance. Methods: Single-cell transcriptomic profiling was performed on samples from a Phase II trial of pembrolizumab (anti-PD-1) in hepatitis B-related HCC patients (NCT03419481). An ICB-resistant HCC mouse model generated via iterative selection was used for validation. The BRD4 inhibitor AZD5153, administered in oral and nanoparticle formulations, was used to suppress BRD4 and TCF4 expression in systemic and endothelial cells. Quantitative immunofluorescence assessed ECs and detected TCF4 and BRD4 levels. Results: scRNA-seq analysis revealed that tumor vessels from ICB non-responders were enriched with disorganized macrovascular-like endothelial cell (MaVEC), which further increased upon ICB. RNA velocity analysis predicted that MaVEC in non-responders likely originate from liver sinusoidal endothelial cell (LEC), indicating LEC-to-MaVEC trans-differentiation during ICB resistance acquisition. TCF4 was identified as the top regulon in MaVEC, with increased activity along the LEC-to-MaVEC trajectory and association with BRD4. Using AZD5153, we observed that TCF4 can be suppressed in ECs, which caused a significant intra-tumoral reversal from MaVEC to LEC phenotypes, resulting in tumor cell apoptosis and regression. Additionally, nanoparticle-delivered EC-targeted AZD5153 achieved similar tumor-suppressive effects at a lower dosage. Conclusions: Our study reveals that the dynamic trans-differentiation from LEC to MaVEC is associated with ICB resistance in HCC, which can be reversed by BRD4 inhibition and possibly mediated by TCF4 transcriptional regulation, targeting their expression in ECs showed improved therapeutic outcomes. Keywords: Hepatocellular carcinoma, Immune-checkpoint blockade, Endothelial cells, Transcriptional network. Acknowledgement: Funded by General Research Fund (14118424) & AstraZeneca Pre-clinical Oncology Research Programme & Li Ka Shing Foundation. Citation Format: Baoyi Yin, Shufen Chen, Zhewen Xiong, Patrick Wong, Zhuo Yu, Stephan Chan, Alfred Sze-Lok Cheng. The role of endothelial cell remodeling in driving immunotherapy resistance of hepatocellular carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3852.

Abstract 2751: Role of transcriptional regulation in chromatin architecture and immunotherapy resistance of hepatocellular carcinoma

Abstract Introduction: Chromatin architecture is frequently altered in cancers, including one of the most prevalent cancers, hepatocellular carcinoma (HCC). Emerging evidence suggests that liquid-liquid phase separation plays a crucial role in shaping chromatin architecture and regulating gene transcription. Immune checkpoint blockade (ICB) therapy has become a new standard of care for HCC. However, the majority of HCC patients exhibit either primary or adaptive resistance to this immunotherapy, highlighting the urgent need for effective combination therapies. This study aims to uncover the differences in chromatin architecture between ICB-sensitive and resistant HCC tumors and dissect the underlying regulatory mechanisms. Methods: Hi-C, ATAC-seq, CUT&Tag, and RNA-seq were performed on RIL175 cells from a well-established adaptive ICB-resistant mouse model to generate the multi-omics dataset. To explore the functional role of F1 in ICB resistance of HCC, RIL175-PD1-R cells with F1 knockdown were constructed using the lenti-virus infection. To explore the phase separation ability of F1, F1 protein tagged with mCherry was expressed in E. coli and purified. Droplet formation assay and fluorescence recovery after photobleaching (FRAP) assay were applied. Results: Three-dimensional genomic analysis by Hi-C reveals comparable chromatin interactions between ICB-sensitive and resistant cells at both compartment and topologically associated domain (TAD) levels but an increase in loops within ICB-resistant cells. In accord with this, ATAC-seq results indicate upregulated chromatin accessibility in a subset of chromatin regions in ICB-resistant cells compared to ICB-sensitive cells. Motif analysis identifies F1 as the top transcription factor enriched for the open chromatin regions. CUT&Tag shows that F1 binds to those accessible regions and alters chromatin looping in ICB-resistant cells. Modulating F1 expression in HCC tumor cells significantly increases the therapeutic efficacy of ICB treatment. We also found that F1 undergoes phase separation, which may mediate the changes in chromatin architecture between ICB-sensitive and resistant cells. Discussion and Conclusion: F1 undergoes phase separation in HCC, particularly in ICB resistant cells. This characteristic of F1 may shape specific three-dimensional chromatin architecture in HCC cells, contributing to their resistance to ICB treatment. Targeting F1 condensates may provide a novel therapeutic strategy to reprogram the genome architecture to improve ICB efficacy for patients with HCC. Keywords: chromatin architecture, phase separation, hepatocellular carcinoma, immune checkpoint blockade (ICB) therapies Acknowledgement: This work is supported by NSFC/RGC (460795048) & Li Ka Shing Foundation. Citation Format: Siyun CHEN, Yingnan LIN, Chengpeng Zhong, Yalin TU, Haoran WU, Yiling ZHANG, Jia WANG, Alfred Sze Lok CHENG. Role of transcriptional regulation in chromatin architecture and immunotherapy resistance of hepatocellular carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2751.

Abstract 3221: Identification and characterization of FAK kinase as a novel regulator of PD-L1 stability and activity in cancer immune evasion

Abstract The PD-1/PD-L1 axis is an essential immune checkpoint that modulates T cell-mediated immune responses. Cancer cells exploit this checkpoint to inhibit anti-tumour immune activity, enabling immune evasion. Thus, blocking PD-1/PD-L1 using antibodies has become a cornerstone of cancer immunotherapies. However, despite promising results, these therapies show limited response rates in many cancer patients and often lead to adverse effects, impacting its overall efficacy. Therefore, understanding of the molecular mechanism underlying immunotherapy resistance is critical for successful cancer immunotherapy. Emerging research indicates that phosphorylation of PD-L1 by kinases significantly influence PD-L1 stability and functionality, thereby contributing to cancer cell immune evasion and influencing immunotherapy outcomes. However, although 560 kinases have been identified in the human kinome, there is no systematic screening of kinases regulating PD-L1 stability and its function in immune evasion. By using a PD-L1 NanoLuc luciferase biosensor, in this study, we performed a systematic kinome-wide screening and identified focal adhesion kinase (FAK) as a novel regulator of PD-L1 stability in cancer. Further investigation of FAK-mediated PD-L1 signalling and its role in immune evasion may provide a new therapeutic strategy for cancer treatment. Citation Format: Asia-lily Boyd, Prem Khanal, Yawei Hao, Xiaolong Yang. Identification and characterization of FAK kinase as a novel regulator of PD-L1 stability and activity in cancer immune evasion [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3221.

Abstract 4399: Systematic interrogation of cell states that drive inflammation tolerance and therapy resistance in cancer

Cancers develop and evolve by overcoming numerous selective pressures including cell-intrinsic growth-inhibitory mechanisms, microenvironmental stresses, and therapeutic interventions. In addition to mutational mechanisms, cancer cell adaptation via transcriptional plasticity remains an important but poorly understood driver of cancer cell fitness. Specifically, when disease progression occurs after targeted therapy treatment, genomically-defined mechanisms of resistance are only identified in approximately 50% of patients, suggesting that non-genomic mechanisms play an equally important role. Thus, there is a pressing need to devise systematic and scalable methods to determine essential drivers of prognostically-relevant cancer cell states, and to engineer state-relevant models that can be applied to discover new therapeutic approaches that overcome these modes of resistance. Here, we present a strategy to systematically identify transcriptional drivers of inflammation- and therapy-resistant cancer cell states. We have successfully established a high-throughput screening pipeline to 1) over-express human transcription factor (TF) isoforms in a pooled format within cancer models; 2) perform positive-selection screens under a variety of selective pressures to define transcriptional mechanisms of inflammation tolerance and therapy resistance; 3) perform single-cell RNA sequencing to define families of TFs that drive specific cell state transitions and resistance phenotypes; and 4) conduct mechanistic studies to examine the downstream signaling networks that facilitate cancer cell growth in the setting of these selective pressures. We uncover TFs and states that drive resistance in a tissue- and perturbation-specific manner, as well as states that are robust to multiple different selective pressures. A subset of TFs and associated gene programs induce tolerance to chronic interferon exposure in our screens and are reflected in vivo in the setting of immunotherapy resistance. We also identify distinct TF families that drive resistance to pharmacologic KRAS inhibition. Notably, cancer cell states that confer resistance to KRAS inhibition also exhibit cross-resistance to other MAPK pathway inhibitors, suggesting that cell state adaption may contribute to multi-drug resistance as cancers progress. More broadly, our strategy enables the systematic identification and mapping of cell states that drive cancer cell fitness to in vivo clinical atlas datasets, providing a framework for interpretation of clinical phenotypes and potential predictive modeling of cell state evolution. Furthermore, the cell state-defined models developed here provide a substrate for new therapeutic discovery using cancer cell states as biomarkers, a key next step to developing combination strategies that overcome therapeutic resistance and improve outcomes for patients. Citation Format: Srivatsan Raghavan, Evelyn Y. Tong, Aswanth H. Mahalingam, Zixin Chen, Jacob Smigiel, Tsukasa Shibue, Andrew Navia, Alex K. Shalek, Lorin Crawford, Ava Amini, Peter S. Winter. Systematic interrogation of cell states that drive inflammation tolerance and therapy resistance in cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4399.

Abstract 4626: A machine learning model for predicting hyperprogressive disease risk in cancer immunotherapy: Integrating clinical and molecular data

Abstract Background: Immunotherapy has revolutionized cancer treatment, offering significant efficacy across various cancer types. However, hyperprogressive disease (HPD)—a severe and unexpected acceleration of tumor growth—poses a critical challenge, as some patients experience this adverse effect. Early and accurate identification of HPD risk is essential for optimizing immunotherapy strategies and improving patient outcomes. To address this, we developed a machine learning-based predictive model for HPD risk across cancer types, integrating clinical and molecular data. Methods: This study analyzed data from 1, 005 patients with solid tumors treated with at least one dose of immune checkpoint inhibitors between 2018 and 2024. After stringent data preprocessing, feature selection was performed using chi-square and Fisher’s exact tests to identify significant variables associated with HPD. Logistic regression evaluated feature importance, accounting for multicollinearity, correlation, and interpretability. A random forest algorithm was then employed to construct a predictive model based on selected features. Results: From an initial pool of 44 variables, eight features were identified as significant predictors of HPD, including HPD-associated genes, immunotherapy resistance genes, cancer type, bone metastasis, the number of metastatic sites, lactate dehydrogenase levels, neutrophil-to-lymphocyte ratio, and treatment regimen. Patients lacking pre-immunotherapy genetic and blood test data for the selected features were excluded, resulting in a final cohort of 363 patients across 17 cancer types, including 52 HPD cases and 311 non-HPD cases. The model demonstrated strong predictive performance, achieving a cross-validated AUC of 0.87, sensitivity of 0.90, and specificity of 0.78. On an independent test set, the model attained an AUC of 0.71, sensitivity of 0.70, specificity of 0.73, and overall accuracy of 73%. Using the same eight features, a traditional logistic regression model was constructed (AUC = 0.66) and found to be less effective than the machine learning model. Conclusions: We identified eight statistically and clinically significant predictors of HPD and developed the first high-sensitivity pan-cancer predictive tool combining molecular and clinical features. This machine learning-based approach enhances HPD risk stratification, assisting clinicians in tailoring immunotherapy and mitigating risks, ultimately improving patient outcomes across cancer types. Citation Format: Haitao Wang, Lili Wang, Yu Zhang, Hongzheng Li, Nan Li, Qijie Jian, Yaxian Wang, Zhijian Song, Fei Pang, Kai Wang. A machine learning model for predicting hyperprogressive disease risk in cancer immunotherapy: Integrating clinical and molecular data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4626.

Abstract 2222: Trimethylamine-N-oxide (TMAO) mediates immunotherapy resistance in advanced HCC

Abstract Recent approvals of immune checkpoint inhibitor (ICI) based regimens have transformed the clinical landscape of advanced hepatocellular carcinoma (HCC), however, clinical responses are only observed in a minority of patients. Trimethylamine-N-oxide (TMAO) is an amine oxide biosynthesized from trimethylamine (TMA), generated when gut bacteria metabolize dietary choline via the enzyme CutC. While prior studies have shown that TMAO concentrations positively correlate with the risk of several gastrointestinal malignancies, newer studies have demonstrated that TMAO may enhance antitumor immunity in some tumor types. TMAO may therefore play an important role in modulating the tumor immune response, but its significance in HCC is unknown. We prospectively collected whole blood from patients with HCC prior to initiating standard-of-care ICI-based therapy at Johns Hopkins (IRB #00267960). Clinical records were reviewed for demographic characteristics and clinical outcomes. Additionally, we studied the impact of acute and chronic choline supplementation on anti-PDL1 efficacy in a syngeneic murine model of HCC (Hep53.4 in C57BL/6 mice). Baseline choline, TMA, and TMAO plasma concentrations in clinical and preclinical specimens were quantified via liquid chromatography-mass spectrometry (LC-MS) utilizing the SCIEX Triple Quad 6500+ system. The clinical cohort included 42 patients with advanced HCC. Patients were predominantly male (81.0%) with Barcelona Clinic Liver Cancer (BCLC) stage C disease (73.8%) who underwent treatment with atezolizumab + bevacizumab (52.4%) or durvalumab + tremelimumab (31.0%). In a multivariable Cox model, higher plasma TMAO concentration at baseline was associated with inferior progression-free survival (PFS; HR 1.36 per interquartile range [IQR] increase, p=0.032) and overall survival (OS; HR 1.63 per IQR increase, p<0.001) after controlling for BCLC stage, albumin-bilirubin index, alpha fetoprotein (AFP) concentration, and baseline choline and TMA concentrations. In our preclinical model, mice given chronic choline supplementation (1% choline diet for 14 days prior to treatment with anti-PDL1 therapy) had increased plasma TMAO concentrations (p<0.001) and exhibited no response to anti-PDL1 therapy, compared to mice given standard and acute choline (1% choline diet starting concurrently with anti-PDL1 treatment) diets. Our clinical and preclinical results suggest that high plasma TMAO concentration is associated with inferior outcomes on ICI therapies in HCC. These findings contrast with existing preclinical studies that suggest TMAO may potentiate responses to ICI therapy. We hypothesize that chronic TMAO exposure may lead to immune cell exhaustion via promotion of an inflammatory microenvironment and oxidative stress. Studies examining the effect of CutC inhibition are ongoing in our laboratory and may provide further rationale for targeting TMAO in HCC. Citation Format: Mari Nakazawa, Vivian L. Raj, NV Rajeshkumar, Noushin Rastkari, Michael Davis, Elizabeth De Oliveria, Madelena Brancati, Ervin Griffin, Kabeer Munjal, James Leatherman, Sarah Mitchell, Marina Baretti, Won Jin Ho, Mark Yarchoan, Chi V. Dang. Trimethylamine-N-oxide (TMAO) mediates immunotherapy resistance in advanced HCC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2222.