Multimodal Analysis Research Articles

Abstract 4631: Deciphering immunotherapy response variability and immune-related toxicities through longitudinal multimodal ctDNA analysis in advanced melanoma

Abstract Background: Immune checkpoint inhibitors (ICIs) have revolutionized treatment of advanced cutaneous melanoma by leveraging the host immune system to target and eliminate tumour cells. However, predictive biomarkers for clinical response to ICIs remain elusive, limiting the ability to stratify patients effectively. The heterogeneity in treatment outcomes, coupled with the limited sensitivity and timeliness of radiographic assessments, particularly in patients with stable disease, underscores the urgent need for molecularly guided, minimally invasive approaches for real-time monitoring of therapeutic efficacy. Biomarkers, such as circulating tumour DNA (ctDNA) and T-cell clonal dynamics, offer promise not only in assessing tumour burden and dynamics but also potentially in providing early indicators of immune-related adverse events (irAEs). Methods: We analysed 70 longitudinal plasma samples from 13 patients with resectable or unresectable, Stage III melanoma from the MELResist study. These patients were treated with either pembrolizumab or a combination of ipilimumab and nivolumab. PD1/PDL1 expression in FFPE tumour samples was quantified using digital pathology, and deep whole-genome sequencing (around 30x coverage) was performed on tumour tissue to profile genomic alterations. By integrating ctDNA analysis (>10x WGS coverage) with matched tumour samples using the cfDNAPro, ichorCNA and INVAR2 pipelines, we conducted serial WGS-based monitoring to evaluate molecular response dynamics as measured by copy number alterations, fragmentomics and patient-specific tumour mutations respectively. Additionally, irAEs were assessed through longitudinal tracking of the peripheral T-cell repertoire (TCR) using TCR-sequencing of buffy coat samples, alongside with plasma protein expression profiling via the Olink Explore panel. Results: In evaluating ctDNA molecular response, we found a copy number fraction ranging from 0 to 42% by ichorCNA analysis, with higher fractions at later timepoints correlating with poorer overall survival. Fragmentomic analysis revealed changes in 4-mer end motif frequencies and the ratio of short-to-long fragments (100-150 bp vs. 151-220 bp). Specifically, a decreased CCCA motif frequency and an increased short-to-long fragment ratio at later timepoints were associated with poorer overall survival. Conclusion: ctDNA dynamics during immunotherapy for advanced melanoma reflect clinical outcomes. Changes in ctDNA can be identified using tumour informed methods with high sensitivity but at additional cost or tumour agnostic methods with lower logistical burden. Together with our assessment of T-cell clonal dynamics and proteomic changes, these approaches will offer valuable insights for monitoring therapeutic benefits and immune-related adverse events (irAEs) during immunotherapy. Citation Format: Zhao Cheng, Sarah McDonald, Hannah Thomas, Ze Zhou, Paulius Mennea, Haichao wang, Dmitry Shcherbo, Christina Zhang, Grainne McAndrew, Fazlur Talukdar, Wendy N. Cooper, Hui Zhao, Nitzan Rosenfeld, Philippa G. Corrie, Amit Roshan. Deciphering immunotherapy response variability and immune-related toxicities through longitudinal multimodal ctDNA analysis in advanced 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 4631.

Abstract 5540: Deconstructing the lethality of recurrence in ovarian cancer through multi-modal spatiotemporal analysis

Abstract Epithelial ovarian cancer (EOC) represents the world’s most lethal gynecological malignancy and is commonly associated with chemo-resistance and recurrence. Strikingly, up to 70% of EOC patients will relapse where subsequent care becomes palliative. Owing to a lack of model systems and a scarcity of matched treatment-naïve and recurrent clinical samples, few studies have addressed the complex relationship between first-onset EOC and its respective relapse. Identification of in situ tumor phenotypes that resisted initial chemotherapy treatment and initiate recurrent tumors, will allow us to understand which cellular features lead to relapse. This will allow us to better predict tumor recurrence and to ultimately suggest novel therapies. Here we analyze primary and recurrent tumor pairs using highly multiplexed imaging technologies. Through integration of spatial transcriptomics and imaging mass cytometry (IMC), we identify and classify resistance phenotypes in over 320 EOC patients, identifying phenotypes that persist in recurrent tumors. We leverage these technologies to assign novel proteo-transcriptomic resistance modules that allow us to independently observe these phenotypes across multiple single-cell modalities. To uncover phenotype-specific drug sensitivities, we utilized scRNA-seq data to predict in silico drug-response and validated these findings through large-scale image-based ex vivo drug screening. These screens revealed significant multi-drug resistance (MDR) properties of persistent phenotypes, but unveiled sensitivity to alternate FDA-approved drug classes. Together, our work represents the first-of-its-kind high-dimensional map of resistant and recurrent EOC, offering tangible therapeutic solutions for the treatment and prevention of recurrence. Citation Format: James M. Whipman, Francis Jacob, Viola Heinzelmann, Bernd Bodenmiller. Deconstructing the lethality of recurrence in ovarian cancer through multi-modal spatiotemporal analysis [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 5540.

Abstract 1026: Multi-modal liquid biopsy analysis of early-stage breast cancer patients

Abstract The 5-year breast cancer survival rate in the United States is 91% overall with a 99% survival rate for localized disease. Early detection via mammograms has largely been responsible for the success of finding breast cancer early; however, mammograms have drastically decreased effectiveness in women with dense breasts tissue which is approximately half of the US population. To better serve these women, an alternative screening method is necessary. Astrin Biosciences has developed a multi-modal liquid biopsy-based alternative for breast cancer screening in this population. Recent literature demonstrates that cancer cells disseminate into blood vessels even at early stages. These early-stage cells are a potential source of late recurrence and therapy resistance but have strong potential benefit in early detection as well as provide opportunities to determine possibly curative interventions. Although these cell assays have a huge potential benefit, they generally leave some to be desired as far as sensitivity due to the rarity of these cells with less than 10 cells per 10 mL blood tube from circulation. To overcome this limitation, Astin’s algorithm pairs our AI-empowered proprietary holographic imaging platform with plasma proteomics. This pairing provides increased sensitivity over a cell-based assay alone while maintaining the ability to detect disseminating cells for further analysis in patients with enough cells present. This combined technology was evaluated in a cohort of more than 100 newly diagnosed early-stage breast cancer patients (Stage 0, 1, and 2). Blood samples were collected prior to initiation of therapy and the plasma and cell fractions were separated. Proteomic analysis was performed on the plasma sample while Astrin’s holographic imaging platform combined with in-flow protein marker expression was used to analyze the samples for circulating tumor cells. Preliminary data from this study demonstrates evidence of early dissemination detectable in the majority of breast cancer patients. GSEA analysis of protein expression revealed well known pathways associated with cancer biology. Astrin Biosciences’ combined assay has the ability to detect cancer in women with dense breast, increasing the likelihood cancer is detected early in this population. Citation Format: Kaylee J. Kamalanathan, Alec Horrmann, Nathaniel R. Bristow, Yash Travadi, Kevin Mallery, Catalina Galeano-Garces, Song Yi Bae, Alexa Hesch, Grant Schaap, Yuliya Olimpiadi, Sarah Pederson, Jiarong Hong, Justin Drake, Jayant Parthasarathy, Badrinath R. Konety. Multi-modal liquid biopsy analysis of early-stage breast cancer patients [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 1026.

Abstract 6476: Multimodal single cell analysis of cholangiocarcinoma subtypes reveals distinct immune microenvironments

Abstract Cholangiocarcinoma (CCA), an aggressive cancer of the bile ducts, is classified into intrahepatic (iCCA), perihilar (pCCA), and distal (dCCA) anatomic subtypes. Each subtype has a distinct epidemiology and pathogenesis. The tumor immune microenvironment (TIME) of each subtype is poorly understood. CCAs are poorly immunogenic tumors and systemic therapy has had subpar efficacy in CCA subtypes. Thus, we hypothesized that each anatomic subtype has a distinct TIME. To address this hypothesis, we completed single cell sequencing of tumor, liver, and bile duct from patients with either iCCA or pCCA/dCCA. 182,054 cells from paired liver (n=16), tumor (n=16) (9 iCCA, 7 pCCA/dCCA), and normal bile duct (n=9) samples were obtained from treatment-naïve patients undergoing resection from June 2020 to August 2023. The tissues were homogenized and flow sorted for immune (CD45+) populations immediately following resection. The cells then underwent single cell RNA sequencing (scRNA-seq) and cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) with antibodies for 157 proteins. Multiple immunocompetent, syngeneic/orthotopic murine models of iCCA with distinct genetic drivers were used to assess immunotherapeutics (PMID: 38458319). Immune cells of interest were isolated by flow sorting and functionally validated via suppression assays and multiparameter flow cytometry. Additionally, a novel pCCA murine model was established by injection of an Pik3caH1047R/p27Kip1-derived extrahepatic biliary cancer cell line into the porta/hilar tissue plane. Compared to non-involved liver, the bile duct immune microenvironment composition closely resembled that of the tumor. Multimodal analysis demonstrated both human and murine iCCA tumors have a significant upregulation of regulatory T cells (Treg) expressing T cell immunoreceptor with Ig and ITIM domain (TIGIT) compared to controls. Suppression assay with intratumoral TIGIT+ Treg from mouse compared with TIGIT- Treg indicate TIGIT+ Treg are more potent in inhibiting proliferation of CD8+ T cells. Moreover, TIGIT blockade using a monoclonal antibody resulted in a significant reduction in tumor burden in multiple iCCA murine models, including SB1, a model resistant to anti-PD-(L)1 monotherapy. Meanwhile, human pCCA tumors had a significant increase in IgA+ plasma cells compared to iCCA tumors. An innovative genetically and anatomically relevant pCCA murine model was developed to interrogate this unique plasma cell population. In summary, these data suggest that iCCA and pCCA have distinct immune microenvironments which likely impacts the differences in response to ICB between the subtypes observed in phase III clinical trials. Additionally, the uninvolved, tumor-adjacent bile duct immune microenvironment resembles that of the tumor, suggesting a pre-malignant niche that facilitates tumor development. Citation Format: Hannah E. Stumpf, Jingchun Yang, Haidong Dong, Rory L. Smoot, Sumera I. Ilyas. Multimodal single cell analysis of cholangiocarcinoma subtypes reveals distinct immune microenvironments [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 6476.

Abstract 777: Revealing the transcriptional and proteomic spatial neighborhoods of early prostate cancer and the tumor microenvironment

Intermediate-risk prostate cancer (PCa) poses a challenge for treatment decisions, made more complex by its heterogeneous nature. Growing evidence highlights the critical role of the tumor microenvironment (TME), in cancer initiation, progression, and response to treatment, underscoring the importance of profiling cancer cells but also the TME within their spatial context. Here, we present single-cell in situ transcriptomic and spatial proteomics results from 300 intermediate-risk PCa prostatectomy samples. Multiple regions of the prostate, including regions of the index lesion, secondary lesions, regions of benign and normal were sampled then analyzed using Bruker’s CosMx SMI platform for a 6000-plex RNA assay in a subset of patients; in addition to proteomic profiling using the GeoMx DSP platform. Clustering assigned cells into 15 cell types based on their transcriptional profiles. Each cell type was tested for significantly different proportions in tumor versus non-tumor samples. The cell type with the greatest bias had a 9-fold greater proportion in tumor samples (adjusted P < 0.01). ERG, largely associated with the TMPRSS2-ERG fusion in PCa, was most specifically expressed in cells of this type. Conversely, the cell type with the greatest bias towards non-tumor samples (3-fold greater proportion, adjusted P < 0.01) is characterized most specifically by the expression of MSMB, which encodes an immunoglobulin binding factor previously found to have decreased expression in PCa. Other evidence of intratumoral heterogeneity was found in samples that are not dominated by these stereotypical cell types. For example, a relatively higher-grade sample from an index tumor possessed high proportion of the tumor-dominant cell type whereas a relatively lower-grade sample from the same tumor did not. Similarly, the tumor and non-tumor samples arising from the same patient showed similar cell type composition, highlighting inter-patient heterogeneity. Moreover, we identified proteins with significantly different expression in tumor versus non-tumor samples. In the TME, EpCAM and CD56 were significantly overexpressed and underexpressed in tumor versus non-tumor regions, respectively (adjusted P < 0.01). In the epithelium, cleaved caspase-9, which initiates the caspase cascade leading to apoptosis, was the most significantly underexpressed protein in tumor samples (adjusted P < 0.01). However, the expression of the mRNA encoding caspase-9 is not highly correlated with the expression of cleaved caspase-9, highlighting the added benefit of protein expression profiling. Taken together, this multi-modal spatial analysis of multiple samples from the same PCa patients facilitates a detailed understanding of early PCa for biomarker discovery and pathways of therapeutic intervention. Citation Format: Anna Y. Lee, Megan Hopkins, Linda Liao, Vida Talebian, Tamara Jamaspishvili, David M. Berman, Melanie Spears, Jane Bayani. Revealing the transcriptional and proteomic spatial neighborhoods of early prostate cancer and the tumor microenvironment [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 777.

Abstract 5317: Multimodal spatial analysis of colon cancer tissue reveals emergence of an immunosuppressive tumor maintenance mechanism

Abstract Background: Colorectal cancer is the second most fatal cancer worldwide, accounting for 9% of all reported cancer deaths, and is the number one cause of death among people under the age of 50. Rapid technical advances in the field of spatial biology are allowing us to gain new insight into colorectal cancer biology and patient heterogeneity, which can help to guide the development of immunotherapies. Using a combination of modalities such as sequential multiplex immunofluorescence (mIF) and spatial transcriptomics (STx) on a colon cancer sample we determined the spatial distribution, phenotype, function, and a cytokine and chemokine profile within the colon cancer tissue. Experimental Procedure: 5 um FFPE sections were prepared from a colon cancer sample exhibiting moderate to poorly differentiated adenocarcinoma of T3, N0 staging. mIF was performed using Lunaphore COMET with a 20-antibody panel, followed by H&E staining and imaging. An adjacent section was used for STx with the 10x Genomics Visium CytAssist platform. Data was analyzed with Lunaphore Horizon®, 10x Genomics Loupe Browser software and PredxBio SpaceIQ™ unbiased cell typing, microdomain discovery and network biology platform that co-analyzes mIF and STx datasets. Results: The tumor showed an aberrant growth pattern and a highly proliferated crypt region. Three distinct microdomains in the tumor tissue were discovered: one with high levels of PD-L1+ cells, a second with high levels of NK cells and a third with low levels of infiltrating cells in the lumen and crypt. There was significant presence of VISTA+ and IDO1+ cells throughout the lumen. These cells interacted with CD4+ and CD8+ cells which were found to be present in both lumen and crypt regions. PD-L1 was prominent in the cells found in both lumen and crypt. Finally, NK cells were abundant throughout the lumen and in regions adjacent to the crypt. Conclusions: Using multimodal spatial imaging and spatial co-analysis of mIF and STx data, we were able to elucidate relationships between different cell types in a complex tumor tissue and assign context to their activities. Despite high levels of NK cells, indicative of a possible active anti-tumor immune response, the tumor has developed an immune evasion mechanism due to the presence of cells expressing high levels of the anti-tumor immune response markers such as PD-L1, IDO1, and VISTA. Based on these results, therapies targeting more than one immune checkpoint protein may yield improved response due to an abundance of the existing NK cells. The transcriptomic profiling reveals a complex relationship between various cytokines secreted by different cell populations which alter the landscape of the tumor microenvironment, facilitating the emergence of an immunosuppressive phenotype. Citation Format: Gaurav Joshi, Nicholas Sciascia, Michael Yang, Brian Falkenstein, Raymond Yan, Shannon Quinn, Brenna Fearey, Junya Yoshioka, Arif Burak Tosun, S. Chakra Chennubhotla, Filippo Pullara, Fumiki Yanagawa. Multimodal spatial analysis of colon cancer tissue reveals emergence of an immunosuppressive tumor maintenance mechanism [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 5317.

Abstract 2391: Comprehensive quantification and annotation of viral gene integration and expression across thousands of human cancer models

Viral infections have been demonstrated to influence host gene expression and drive tumorigenesis. This study aims to systematically annotate genome integration and expression of ∼3000 exogenous and human-endogenous viral species across all public DepMap cancer cell lines, representing a diverse set of cancer types and lineages. Our preliminary results from this analysis pipeline successfully recapitulated known Human Papillomavirus (HpV strains 16, 18, 30 and 53) signatures in 12 cervical, 2 head and neck, and 4 engineered prostate cancer cell lines, demonstrating sensitivity at both the strain and gene levels. Associating these with genome-wide CRISPR screening data, we identified dependencies that were more correlated with viral presence rather than their respective lineages. Additionally, we observed expression of Epstein-Barr Virus(EBV) in multiple cancer types, a majority of which were of lymphoid lineage. We also found expression of the env gene of Human Endogenous Retrovirus-K (HERV-K113) in multiple lineages, with the highest average expression in skin cancers. By analyzing whole-genome sequencing we seek to identify viral integration sites in the cell lines’ genomes, to complement viral gene expression. From this large-scale multimodal analysis of features derived from genomic and transcriptomic data combined with gene dependencies, we aim to uncover potential viral contributions to oncogenesis and biomarker or target discovery. This will provide a foundational resource for understanding the role of viruses in cancer, offering new insights into viral-driven mechanisms of disease. Citation Format: Sowmya Iyer, Moore van Tienen, Phillipp Trollman, Yao He, William R. Sellers, Catarina D. Campbell. Comprehensive quantification and annotation of viral gene integration and expression across thousands of human cancer models [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 2391.

Abstract 4637: Nanomechanical signatures: novel class of clinically translatable biomarkers to predict response to immunotherapy in advanced solid tumors

Abstract Background: Advancements in immunotherapy have opened new avenues for the treatment of solid tumors. However response remains poor, especially in case of advanced tumors where resistance to immunotherapy after initial response remains a significant challenge. In recent years it has become evident that mechanical properties of tumor and immune cells as well as of the tumor microenvironment (TME) pose an upstream challenge to efficacy of a variety of therapeutics, including immunotherapy. Approaches such as pretreatment with low-dose radiation therapy (LD-RT) to remodulate the physical tumor microenvironment in combination have shown promise in clinics, reporting as effective as fully extinguishing detectable tumors in patients with previously stage IV tumors in lung and liver. Accurate prediction of benefit by such strategies is of outmost importance. In this work, we report on a novel class of predictive, tumor-agnostic biomarkers for combination immunotherapies: nanomechanical signatures. Methods: To characterize the nanomechanical signature, the clinically-embedded, tissue-preserving AFM-based Automated and Reliable Tissue Diagnostics (ARTIDIS) platform is used. ARTIDIS multimodal analysis includes integration with available histopathological and clinical data. Two preclinical models are used: PD1-resistant lung adenocarcinoma on 129Sv/Ev mice, and gastric cancer on NSG mice, treated with anti-CTLA and anti-PD1 in combination with LD-XRT and anti-mesothelin CAR-T cell therapy in combination with LD-XRT, respectively. Case reports from patients with advanced metastatic solid cancers resistant to immunotherapy enrolled under the on-going ARTIDIS Umbrella protocol at MDACC are also included. Results: Our preclinical studies identify the nanomechanical signature of response to low-dose radiation and immunotherapy with high sensitivity, specificity and AUC (90%-99%). Furthermore, we can identify the unique nanomechanical signature of T cell infiltration. Our clinical studies confirm that a unique nanomechanical signature reflects TME remodulation in different therapeutic regiments among similar clinical and histopathological cases. Conclusions. In this work we demonstrated across different therapeutic strategies, preclinical tumor models and clinical cases that the nanomechanical signature can quantitatively describe the TME remodulation upon LD-XRT that mediates restored response to immunotherapy. This work supports the use of nanomechanical signature as predictive biomarker of response to immunotherapy to be used to support clinical decision making. Citation Format: Sara Nizzero, Nahum Puebla-Osorio, Gitika Srivastava, Reinier Oropesa Nuñez, Kathleen Graham, Papa Diogop Ndiaye, Joylise Mitchell, Ailing Huang, Thomas Riad, Carolina Ortiz-Velez, Sienna Welsh, Philipp Oertle, Tobias Appenzeller, Vittorio Cristini, Marko Loparic, Marija Plodinec, James W. Welsh. Nanomechanical signatures: novel class of clinically translatable biomarkers to predict response to immunotherapy in advanced 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 4637.

Abstract 2178: Multimodal analysis of the PDAC tumor microenvironment in response to radiation and immunotherapy reveals insights into treatment resistance

Abstract Pancreatic ductal adenocarcinoma (PDAC) is a deadly disease with few treatment options and poor prognosis. Immunotherapy is effective in other cancers but shows worse outcomes in PDAC due to a desmoplastic stroma, an abundance of immunosuppressive cells, and a lack of antigen-presenting dendritic cells, which result in low immune infiltration. We previously reported a study combining dual immune checkpoint blockade (ICB) and radiation therapy in metastatic PDAC patients who progressed on at least one line of chemotherapy and showed an 18% overall response rate (ORR) and 29% disease control rate (DCR). Here, we performed a follow-up phase 2 study on previously treated PDAC patients and observed 3.6% ORR, 10.7% DCR, 2.3 months median progression-free survival, and one complete response. To uncover mechanisms of treatment resistance, we profiled 33 samples from 23 patients with 13 paired pre and on treatment biopsies across distinct tissue sites including liver metastases and primary pancreas. Using single nucleus sequencing (>240k cells), single-cell RNA sequencing (>160k cells) with TCR (scTCR) sequencing (>25k cells), and Visium spatial transcriptomics of a subset of samples (n=15), we comprehensively characterized the cell states and gene programs in the TME of metastatic PDAC prior to and after radiation and ICB. Post radiation, tumor cells were enriched for classical states, exhausted CD8 and proliferating T cells were expanded, and interferon-related programs were upregulated in epithelial subtypes, C1QC+ and MHCII+ macrophages, and CD14+/CD16+ monocytes, suggesting cell type-specific responses to treatment. Expanded T cell clonotypes showed phenotypic shifts toward CD8 exhausted and cytotoxic activated states. Using a published gene signature, we identified tumor-reactive T cells, which correlated with known proxies for tumor reactivity including CXCL13 expression. Only 2.2% of T cells were tumor-reactive, and 71% of these came from the single patient who displayed complete response to treatment. Moreover, bulk TCR sequencing of serial peripheral blood samples showed long-term persistence of tumor-reactive clonotypes in this patient over a year post-treatment. Spatial transcriptomics demonstrated co-localization of CXCL13+ T cell, dendritic cell, and inflammatory chemokine gene programs, and separate co-localization of macrophage, CCL8+ CAF, Th17/Tc17, Class II MHC antigen presentation, and interferon stimulated gene programs. Together, our multimodal dataset offers valuable insight into radiation and immunotherapy-induced shifts in the TME. Citation Format: Sarah Kang, Aparna R. Parikh, Milan Parikh, Julie L. Koenig, Leon Pappas, Moshe Sade-Feldman, Lynn Bi, Nicole Carzo, Tarin M. Grillo, Islam Baiev, Olanike Asupoto, Irena Gushterova, Tom LaSalle, Anna Gonye, David P. Ryan, David T. Ting, Dana Pe'er, Theodore S. Hong, Nir Hacohen, Ryan J. Park, Arnav Mehta. Multimodal analysis of the PDAC tumor microenvironment in response to radiation and immunotherapy reveals insights into treatment 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 2178.

“Are You Still Alive?”: Survival Politics, Art, and Grief in Visualizing Palestine

This research analyses the difficulties and rights violations of the Palestinian people through visual narratives. The Palestinian struggle for freedom has been ongoing for many years, especially under Israel’s military rule and settlement policies. The study focuses on the Visualizing Palestine (VP) platform, which aims to strengthen this struggle and develop an alternative visual language against hegemonic media narratives. The population of this study includes all visual narratives and reports produced on the Palestinian issue, including digital platforms, publications and media content. The sample is specifically selected from VP’s 2021-2023 annual reports and 2023-2025 strategy reports. Using techniques such as Semiotic Analysis, visual content analysis and structural analysis, along with Multimodal Discourse Analysis (MDA) based on Kress and Van Leeuwen’s (2006) methodology, this study explores how the visual content of these reports acquires meaning in social, cultural and political contexts. The findings reveal how VP’s visual strategies have reshaped Palestinian identity and historical memory while increasing global awareness and empathy towards the Palestinian issue. The results show that through visual narratives, Palestine’s international visibility is enhanced, awareness of rights violations is strengthened, and global social responses are mobilised. The findings also reveal how visual tools can challenge dominant narratives, promote activism around social justice issues and contribute to the transformation of public discourse.

“Unjust and unworthy portrayal”: a multimodal content analysis of Black experiences in U.S. history textbooks

ABSTRACT Purpose History textbooks remain in wide use globally, and the modes of communication within textbooks have become increasingly multimodal. Therefore, meaning making is not only extended by the inclusion of more visual images, it encompasses the interplay among various modes of representation, including visual images, textual features, and design elements. In this study, we examined how these three modes of representation interact to portray Black experiences in two widely-used, secondary U.S. history textbooks. Research Methods/Approachs Utilizing multimodal content analysis paired with social semiotics and critical race theory–with a focus on whiteness as property, we analyzed 148 multimodal ensembles across two U.S. history textbooks. Findings We found that Black experiences are portrayed (a) through violent depictions that shield white accountability, (b) through sanitized narratives that protect whiteness, and (c) with limited depth and complexity. Implications We address the implications of our findings and offer recommendations to both teachers and teacher educators that highlight the importance of combining the use of an evidenced-based framework for teaching Black history and critical multimodal literacy in social studies classrooms around the world.

Navigating the Multiverse: A Hitchhiker’s Guide to Selecting Harmonisation Methods for Multimodal Biomedical Data

Abstract Introduction The application of machine learning (ML) techniques in predictive modelling has greatly advanced our comprehension of biological systems. There is a notable shift in the trend towards integration methods that specifically target the simultaneous analysis of multiple modes or types of data, showcasing superior results compared to individual analyses. Despite the availability of diverse ML architectures for researchers interested in embracing a multimodal approach, the current literature lacks a comprehensive taxonomy that includes the pros and cons of these methods to guide the entire process. Closing this gap is imperative, necessitating the creation of a robust framework. This framework should not only categorise the diverse ML architectures suitable for multimodal analysis but also offer insights into their respective advantages and limitations. Additionally, such a framework can serve as a valuable guide for selecting an appropriate workflow for multimodal analysis. This comprehensive taxonomy would provide a clear guidance and support informed decision-making within the progressively intricate landscape of biomedical and clinical data analysis. This is an essential step towards advancing personalised medicine. Objective The aims of the work are to comprehensively study and describe the harmonisation processes that are performed and reported in the literature and present a working guide that would enable planning and selecting an appropriate integrative model. Results We present harmonisation as a dual process of representation and integration, each with multiple methods and categories. The taxonomy of the various representation and integration methods are classified into six broad categories and detailed with the advantages, disadvantages and examples. A guide flowchart describing the step-by-step processes that are needed to adopt a multimodal approach is also presented along with examples and references. Conclusions This review provides a thorough taxonomy of methods for harmonising multimodal data and introduces a foundational 10-step guide for newcomers to implement a multimodal workflow.

Coprolite Multimodal Analysis: A Tool for Hyaenid Predator Identification

Paleontologists and archeologists reconstruct ancient ecosystems using data from carnivores’ food remains. Carnivores have evolved to employ two primary feeding strategies: consuming mostly meat and focusing on both meat and bones, and these strategies result in the production of different feces. Hyenas are exemplary meat-eaters and bone-crushers. While fecal characteristics like shape, color, size, and inclusions are often used for species identification, the detailed composition of hyena feces remains largely unexplored. To address this, we conducted a multimodal analysis of feces-like coprolites from four modern Hyaenid species, using scanning electron microscopy (SEM) and Fourier transform infrared spectrometry (FTIR). This approach allowed for the detection and quantification of the proportions of calcium phosphate/carbonate, silts, organic matter, and crystallinity in the coprolites. Our preliminary findings suggest that multivariate statistical analysis of these components could provide a reliable method for species identification based solely on fecal content, results which can be applied in research on fossil materials.

Gaze Direction as Framing for Interaction in Sign Language-Interpreted Routine Postnatal Visits.

This study aims to gain insight into participants' use of gaze in sign language-interpreted consultations in health centres for babies and toddlers in the Norwegian context. The study investigates how gaze direction and eye contact frame the interaction between participants comprising mothers who are deaf and health personnel who are hearing. The empirical material is based on five video recordings of interpreted health consultations, and multimodal interaction analysis is used to examine the participants' use of gaze in these interactions. The analysis demonstrates differences in eye contact that provide participants with access to different footings because of their language modalities and language knowledge. The mothers and the sign language interpreters take responsibility for the interaction when the health professional is unaware of the importance of a specific gaze in sign language. When taking responsibility, they shift between frontstage and backstage activity. In addition, the mothers do not have full access to the interaction, even with sign language interpreters present. The findings show that it is imperative that health personnel know how to communicate with deaf parents and how to work with sign language interpreters.

Energy-resolved neutron imaging and diffraction including grain orientation mapping using event camera technology

Time-of-flight neutron diffraction and energy-resolved imaging each provide unique perspectives into material properties. Neutron diffraction is useful for assessing microstructural parameters such as phase composition, texture, and dislocation densities, though it typically provides averaged data over the sampled volume. Energy-resolved imaging, on the other hand, offers both spatial and spectral information by detecting Bragg edges and neutron absorption resonances, which enables detailed mapping of microstructure and isotopic composition. When combined, these techniques have the potential to enrich our understanding of material behavior across different scales, enhancing our understanding of complex materials. Traditionally, these modalities are conducted on separate instruments, which is time-consuming and poses challenges for data integration. Here, we report the integration of the LumaCam, an event-mode energy-resolved neutron imaging camera with the HIPPO time-of-flight diffractometer at LANSCE. This integration enables simultaneous diffraction and imaging across the full spectrum, with analysis optimized for diffraction and Bragg-edge imaging in the thermal range (0.45–10 Å) and resonance imaging in the epithermal range (0.5–3000 eV), facilitating comprehensive multi-modal analysis. We demonstrate its capabilities through case studies, including spatial mapping of grain orientations in a steel sample and accurate thickness estimations for irregular samples including a depleted uranium cylinder and a natural silver-containing mineral specimen. The combined setup enhances real-time sample alignment and provides comprehensive data for crystal structure, texture, and isotopic composition analysis. This approach opens new possibilities for advanced applications in nuclear engineering, archaeology, and materials science.

Multimodal Critical Discourse Analysis of the Evolution of Female Representation in Advertising

This study examines the evolution of female representation in Chinese advertising, focusing on the shared characteristics of female images across different types of advertisements. Utilizing a multimodal discourse analysis framework, the study explores the distinctive features and transformations of female representation in Chinese advertisements. From a critical discourse perspective, it scrutinizes gender discrimination embedded in advertising discourse and seeks to eliminate bias and discrimination against women in advertising representations. Furthermore, the study investigates how the shaping of female images in advertising has evolved with the development of the internet, highlighting its role in promoting gender equality awareness within the advertising industry.

AI-Driven Early Detection of Cognitive Decline

Early detection of cognitive decline is critical for initiating timely interventions that can delay or mitigate the progression of neurodegenerative disorders such as Alzheimer’s disease. Traditional diagnostic methods, which rely on episodic clinical evaluations and subjective assessments, often miss the subtle, early signs of impairment. Recent advancements in Artificial Intelligence (AI) offer transformative potential by enabling continuous, objective, and highly sensitive analysis of behavioral, linguistic, physiological, and neurological data. This paper explores the role of AI in revolutionizing the early diagnosis of cognitive decline through the integration of machine learning models, natural language processing, and multimodal data analysis. We examine the key data sources—including speech patterns, gait analysis, neuroimaging, and digital biomarkers—that power AI-driven systems and highlight how these tools surpass conventional approaches in accuracy and scalability. The discussion extends to various AI models such as deep learning and ensemble methods, which can detect subtle patterns indicative of mild cognitive impairment before it becomes clinically apparent. Benefits such as personalized monitoring, remote accessibility, and population-wide screening are considered alongside critical challenges, including data privacy, algorithmic bias, and clinical integration. Ethical considerations and future research directions are emphasized, focusing on the need for transparency, inclusivity, and cross-disciplinary collaboration. By showcasing real-world applications and current limitations, this study provides a comprehensive overview of how AI can support earlier diagnoses and improved care for individuals at risk of cognitive decline. The findings underscore AI's promise as an essential tool in the future of cognitive health.

Children’s perceptions of the indoor environment of early childhood education: following children’s photographic voices

ABSTRACT This article highlights some of the results from a four-year study of the indoor environment in nine preschools in three Swedish municipalities. The data in focus concerns children’s perspectives and perceptions of both new and more traditional education settings. The researcher’s lens is directed towards following the children’s photographic voices (Magnusson, Lena O 2021. “Visual Research Material and Diffractive Readings – A Relational Research Story.” International Journal of Qualitative Studies in Education 34 (3): 183–196. https://doi.org/10.1080/09518398.2020.1735564; Magnusson, Lena O, and Annika Åkerblom. 2022. “Children Taking Place in the Institutional Practice Thought Their Photographic Voice.” In Local Childhood in Global Times, edited by Susanne Garvis, and Anette Hellman, 82–102. INTELLECT LTD) while using digital cameras and what children’s agency (Corsaro, William A. 2018. The Sociology of Childhood. Los Angeles: SAGE) and their multimodal (Jewitt, Carey. 2014. The Routledge Handbook of Multimodal Analysis. Abingdon: Routledge; Kress, Gunther R. 2010. Multimodality: A Social Semiotic Approach to Contemporary Communication.) expressions as part of the photographic events can tell us about their interests and place-based engagement. Using thematic analysis, the children’s photographic voices point to four themes: educational practice and everyday places, highlighted people, objects and places, belonging and what if. All these themes point to diverse aspects of what the children’s perceptions of the indoor environment are or can be.

MSAmba: Exploring Multimodal Sentiment Analysis with State Space Models

Multimodal sentiment analysis, which learns a model to process multiple modalities simultaneously and predict a sentiment value, is an important area of affective computing. Modeling sequential intra-modal information and enhancing cross-modal interactions are crucial to multimodal sentiment analysis. In this paper, we propose MSAmba, a novel hybrid Mamba-based architecture for multimodal sentiment analysis, consisting of two core blocks: Intra-Modal Sequential Mamba (ISM) block and Cross-Modal Hybrid Mamba (CHM) block, to comprehensively address the above-mentioned challenges with hybrid state space models. Firstly, the ISM block models the sequential information within each modality in a bi-directional manner with the assistance of global information. Subsequently, the CHM blocks explicitly model centralized cross-modal interaction with a hybrid combination of Mamba and attention mechanism to facilitate information fusion across modalities. Finally, joint learning of the intra-modal tokens and cross-modal tokens is utilized to predict the sentiment values. This paper serves as one of the pioneering works to unravel the outstanding performances and great research potential of Mamba-based methods in the task of multimodal sentiment analysis. Experiments on CMU-MOSI, CMU-MOSEI and CH-SIMS demonstrate the superior performance of the proposed MSAmba over prior Transformer-based and CNN-based methods.

Aspect Enhancement and Text Simplification in Multimodal Aspect-Based Sentiment Analysis for Multi-Aspect and Multi-Sentiment Scenarios

Multimodal Aspect-Based Sentiment Analysis (MABSA) plays a pivotal role in the advancement of sentiment analysis technology. Although current methods strive to integrate multimodal information to enhance the performance of sentiment analysis, they still face two critical challenges when dealing with multi-aspect and multi-sentiment data: i) the importance of aspect terms within multimodal data is often overlooked, and ii) models fail to accurately associate specific aspect terms with corresponding sentiment words in multi-aspect and multi-sentiment sentences. To tackle these problems, we propose a novel multimodal aspect-based sentiment analysis method that combines Aspect Enhancement and Text Simplification (AETS). Specifically, we develop an aspect enhancement module that boosts the ability of model to discern relevant aspect terms. Concurrently, we employ text simplification module to simplify and restructure multi-aspect and multi-sentiment texts, accurately capturing aspects and their corresponding sentiments while reducing irrelevant information. Leveraging this method, we perform three tasks including multimodal aspect term extraction, multimodal aspect sentiment classification, and joint multimodal aspect-based sentiment analysis. Experimental results indicate that our proposed AETS model achieved state-of-the-art performance on two benchmark datasets.