Citations

Abstract Neuroblastoma (NB) accounts for approximately 10% of pediatric cancer-related deaths. High-risk NB is often metastatic at diagnosis with a poor 5-year survival rate (<50%) despite intensive multimodal therapy. GD2, a disialoganglioside glycolipid expressed in nearly all NBs, is a well-validated immunotherapy target. Although GD2-CAR-T cells have been clinically evaluated, their efficacy remains limited due to poor tumor infiltration and the immunosuppressive tumor microenvironment (TME), underscoring the need for alternative immune-based approaches. Given the inherent capacity of macrophages to infiltrate solid tumors, phagocytose cancer cells, and modulate the TME, we engineered the GD2-CAR macrophage (GD2-CAR-M) as a novel therapeutic platform for NB. GD2-specific CAR constructs containing the Hu3F8 single-chain variable fragment (scFv) were designed, validated, and introduced into multiple human and murine monocyte/macrophage cell lines via lentiviral transduction. Fluorescence microscopy and flow cytometry demonstrated a significant enhancement in the phagocytosis of GD2-high NB cells and the phagocytic efficiency correlating positively with GD2 expression levels, suggesting antigen-specific phagocytosis by GD2-CAR-M. Our further in vivo studies using an immunocompetent MYCN-driven NB non-genetically engineered mouse model (MYCN-NGEMM) showed that treatment with GD2-CAR-M markedly reduced tumor burden and prolonged survival compared with controls. Notably, mice bearing GD2-high-tumors exhibited significantly improved survival, including one complete remission among ten treated animals. Histological analyses revealed increased connective tissue deposition and decreased tumor cell density, suggesting TME remodeling. Single-cell RNA sequencing and immunohistochemistry further demonstrated enhanced infiltration of macrophages, neutrophils, and T cells, indicating a shift from a “cold” to a “hot” immune TME following GD2-CAR-M treatment. Consistent antitumor effects were also observed in our dbh-MYCN transgenic zebrafish with spontaneous development of NB following GD2-CAR-M administration. Taken together, GD2-CAR-M exhibits potent antigen-specific antitumor activity and promotes favorable TME remodeling in NB models. These findings highlight GD2-CAR-M as a promising translational immunotherapy strategy for high-risk NB. Citation Format: Shizhen (Jane) Zhu, Ke-En Tan, Kok Siong Yeo, Cheng Zhang, Caleb T. Baker, Alexis M. Sosic, Eddie Lim . Yat Yuen, Choong Yong Ung, Stephanie F. Polites, Yi Lin, William A. Weiss, Hu Li. Harnessing GD2-CAR macrophages to target high-risk neuroblastoma [abstract]. In: Proceedings of the AACR Immuno-Oncology Conference (AACR IO): Discovery and Innovation in Cancer Immunology: Revolutionizing Treatment through Immunotherapy; 2026 Feb 18-21; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Immunol Res 2026;14(2 Suppl):Abstract nr B005.

Abstract Histones variant composition and post-translational modifications (PTMs) of core and linker histones jointly orchestrate chromatin architecture and tissue-specific gene regulation. However, capturing the full complexity of the “histone code” in native human tissues remains challenging. Here we present a human cardiac histone proteoform atlas—comprising intact histone variants with combinatorial PTMs—enabled by a streamlined top‑down proteomics workflow. In a single one‑dimensional reversed‑phase liquid chromatography (LC)-mass spectrometry (MS) run, we achieve baseline separation of all four core histone families (H2A, H2B, H3, and H4) together with the linker histone H1 directly from human myocardium. Notably, this intact-protein analysis preserves the connectivity of co-occurring PTMs to individual histone molecules to reveal the combinatorial histone code, enabling proteoform‑level quantification of histone‑variant composition and PTM stoichiometry at the chromatographic scale. Targeted MS/MS directly resolves co-occurring PTM combinations and distinguishes highly homologous histone isoforms. Using this approach, we assemble the first comprehensive cardiac histone proteoform atlas and uncover previously unobserved histone proteoforms. More broadly, this top-down proteomics workflow provides a robust framework for proteoform-resolved analysis of histone variants and PTMs in complex biological systems.

Abstract Generative artificial intelligence (AI) is rapidly populating medical records with synthetic or partially AI-generated content, creating a feedback loop where future models are increasingly at risk of training on uncurated AI-generated data. However, the clinical consequences of this AI-generated data contamination remain unexplored. Here, we show that in the absence of mandatory human verification, this self-referential cycle drives a rapid erosion of pathological variability and diagnostic reliability of medical data at population scale. By analysing more than 800,000 synthetic data points across clinical text generation, vision–language reporting, and medical image synthesis, we find that models progressively converge toward generic phenotypes regardless of the model architectures. Specifically, rare but critical findings, including pneumothorax and effusions, vanish from the synthetic content generated by AI models, while demographic representations skew heavily toward middle-aged male phenotypes. Crucially, this degradation is masked by false diagnostic confidence. Models continue to issue reassuring reports while failing to detect life-threatening pathology, with false reassurance rates tripling to 40%. Blinded physician evaluation confirms that this decoupling of confidence and accuracy renders AI-generated documentation clinically useless after just two generations. We systematically evaluate three mitigation strategies that can be easily integrated into existing clinical workflows, finding that while synthetic volume scaling fails to prevent collapse, mixing real data with quality-aware filtering effectively preserves diversity. Ultimately, our results suggest that without policy-mandated human oversight, the deployment of generative AI threatens to degrade the very healthcare data ecosystems it relies upon.

Background and Aims: Cholangiocarcinoma (CCA) is a rare but aggressive malignancy with poorly understood genetic susceptibility. To date, genome-wide association studies (GWAS) investigating germline variants associated with CCA risk remain limited. We aimed to identify genetic risk loci for CCA and its clinical subtypes through comprehensive GWAS and post-GWAS analyses. Approach and Results: We conducted a GWAS of 2366 CCA cases and 11,750 controls of European ancestry. Genome-wide significant loci ( p <5×10 −8 ) were identified and further examined through fine-mapping, functional annotation, and HLA imputation. Subgroup analyses were conducted by CCA subtypes and primary sclerosing cholangitis (PSC) status. Cross-trait linkage disequilibrium score regression and Mendelian randomization were employed to investigate the shared genetic architecture and potential causal relationships with a diverse range of traits. We identified 1 new genome-wide significant variant, rs535777 (OR=1.44), near HLA-DRB1/DQA1 associated with CCA, and 2 variants associated with extrahepatic CCA: rs116224263 (OR=0.17) in LINC02506 at 4p15.1 and rs6914950 (OR=1.63) near HLA-DRB1/DQB1 . Stratified analyses revealed rs2395184 (OR=3.51) near HLA-DRA/DRB5 associated with PSC-related CCA, and rs142674434 (OR=2.98) in THSD7A at 7p21.3 associated with non-PSC-related CCA. HLA imputation uncovered new amino acid residues associated with disease risk. Cross-trait analyses identified shared genetic signals between CCA and anthropometric, lipidemic, lifestyle, and medical traits. Mendelian randomization supported putative causal associations for 12 traits with CCA or its subtypes. Conclusions: Our large-scale GWAS highlights new genetic variants and HLA-linked mechanisms underlying CCA susceptibility. Integrating multi-step post-GWAS approaches enhances understanding of CCA pathogenesis and may facilitate the development of risk biomarkers for early detection and precision prevention strategies.

ABSTRACT Objective Identifying risk factors enables stratification of patients’ susceptibility to inflammatory arthritis immune-related adverse events (IA-irAE). This retrospective study examines whether preexisting osteoarthritis (OA) increases the likelihood of de novo IA in patients treated with immune checkpoint inhibitors (ICIs). Methods The prevalence of OA among ICI–treated patients who developed IA-irAE, those who developed other types of irAEs but not IA (non-IA irAE), and those who did not develop any irAEs (non-irAE) were compared. Electronic medical records were reviewed to extract demographic, clinical and laboratory data. Group comparisons and logistic regression analyses were performed. Results 181 de novo IA-irAE patients, 140 non-IA irAE patients and 170 non-irAE patients were included. The prevalence of OA was significantly higher in the IA-irAE group (69%) than the non-IA irAE group (48%) and the non-irAE group (48%) (both p < 0.001). The IA-irAE group demonstrated a higher frequency of multisite OA, with predominant hand involvement (62%) than the non-IA irAE with OA group (13%) and the non-irAE with OA group (13%) (both p < 0.001). A family history of autoimmune disease (AID) (OR 2.03, 95% CI 1.02–4.05), preexisting OA (OR 2.88, 95% CI 1.85–4.52) and melanoma (OR 2.63, 95% CI 1.56–4.47) were identified as risk factors for the development of IA-irAE. Conclusions OA was more prevalent among ICI-treated patients developing IA-irAE than those who did not. Hand OA was the most common OA pattern in IA-irAE patients. Preexisting OA, melanoma and a family history of AID were risk factors for IA-irAE.

Abstract Deep learning models enable the prediction of clinical endpoints from whole-slide images (WSIs), but many such models function as “black boxes”, lacking transparency about whether and which histomorphological patterns drive their predictions, hindering interpretability and clinical adoption. Here we propose a human-in-the-loop explanation framework, MorphoXAI, which provides both local and global interpretability for deep learning models by incorporating human-expert interpretations. At the global level, it reveals the histomorphological patterns on which the model consistently relies to distinguish between classes of WSIs, as well as the patterns associated with confusion between classes. At the local level, it indicates which of these patterns are used in the prediction of an individual WSI and which regions within the slide correspond to such patterns. We validated our method on a deep learning model trained for ovarian tumor histologic subtype prediction. The results show that our framework generates explanations that accurately reflect the histomorphology underlying the model’s predictions at both global and local levels. For interpretability and clinical utility in diagnostic contexts, human evaluation results showed that our explanations were easy to interpret, rich in diagnostic features, and directly helpful for diagnostic decision-making, thereby enhancing pathologist-AI collaboration. Our work highlights that unifying global and local explanations and grounding them in expert-interpreted morphology enhances the interpretability and verifiability of deep learning models, thereby facilitating the transparent deployment of such models in clinical practice.

Objective: This study aims to develop a contrast-free, high-sensitivity ultrasound method, denoted as Radon transform-based flow measurement (R-Flow), for in vivo mapping of microvascular flow vectors and for establishing R-Flow-derived vector-field metrics to noninvasively quantify microcirculatory patterns in liver cirrhosis. Impact Statement: R-Flow enables robust, contrast-free imaging of microvascular dynamics and demonstrates translational feasibility in the human liver. Its direction-aware indices offer pilot in vivo quantification of flow redistribution and remodeling, providing unique insights into hepatic flow dynamics. Introduction: Microvascular dysfunction is a hallmark of many diseases, yet noninvasive visualization and quantitative assessment of abnormal microcirculation remain limited. Methods: R-Flow leverages Radon transform to decode red blood cell dynamics from the spatiotemporal domain of ultrasound flow signals, reconstructing velocity vectors at microvascular scale. From these vector maps, unique direction-aware indices are further derived to characterize flow distribution and heterogeneity. Results: Validated across simulations, phantoms, and in vivo studies, R-Flow provides robust velocity estimation across a wide range (1 to 60 mm/s). Notably, it enables high-sensitivity microvascular flow vector mapping of human liver, showing strong agreement with references ( r > 0.9). In a rat model, these direction-aware indices revealed a shift from healthy multipath perfusion in control livers to directionally biased vascular pattern in cirrhotic livers, demonstrating significant correlation with pathological indicators. Conclusion: R-Flow enables noninvasive, contrast-free mapping of microvascular blood flow velocity and offers a promising approach for high-resolution assessment of microvascular flow characteristics.

INTRODUCTION: This study used the novel C-reactive protein-albumin-lymphocyte (CALLY) index to explore its relationship with the risk of developing gallstones. METHODS: This study conducted a cross-sectional analysis of the National Health and Nutrition Examination Survey data from 2017 to 2020; multivariable logistic regression examined the CALLY-gallstone association. Restricted cubic splines tested nonlinearity. Subgroup and mediation analyses explored population variations and mediating effects (Conicity, lipid accumulation product, and a body shape index). Receiver operating characteristic curves compared CALLY, systemic immune inflammation index, systemic inflammatory response index, and aggregate index of systemic inflammation predictive performance. RESULTS: After comprehensive adjustments, the highest CALLY quartile had significantly lower gallstone risk vs the lowest (odds ratio 0.56, 95% confidence interval 0.34–0.94). Restricted cubic splines revealed a linear inverse correlation. Subgroup analyses generally supported this inverse relationship. Mediation analysis identified the Conicity index as the strongest mediator (21.47%, P < 0.001), followed by lipid accumulation product and a body shape index (10.72% each, P < 0.001). Receiver operating characteristic analysis showed CALLY (area under the curve = 0.604) outperformed systemic immune inflammation index (0.540), systemic inflammatory response index (0.550), and aggregate index of systemic inflammation (0.546). DISCUSSION: A significant inverse association exists between the CALLY index and gallstone prevalence. The CALLY index demonstrates superior predictive ability compared with other indices, suggesting its potential utility as an objective biomarker for early gallstone risk identification.

Abstract Generative artificial intelligence (AI) is rapidly populating medical records with synthetic or partially AI-generated content, creating a feedback loop where future models are increasingly at risk of training on uncurated AI-generated data. However, the clinical consequences of this AI-generated data contamination remain unexplored. Here, we show that in the absence of mandatory human verification, this self-referential cycle drives a rapid erosion of pathological variability and diagnostic reliability of medical data at population scale. By analysing more than 800,000 synthetic data points across clinical text generation, vision–language reporting, and medical image synthesis, we find that models progressively converge toward generic phenotypes regardless of the model architectures. Specifically, rare but critical findings, including pneumothorax and effusions, vanish from the synthetic content generated by AI models, while demographic representations skew heavily toward middle-aged male phenotypes. Crucially, this degradation is masked by false diagnostic confidence. Models continue to issue reassuring reports while failing to detect life-threatening pathology, with false reassurance rates tripling to 40%. Blinded physician evaluation confirms that this decoupling of confidence and accuracy renders AI-generated documentation clinically useless after just two generations. We systematically evaluate three mitigation strategies that can be easily integrated into existing clinical workflows, finding that while synthetic volume scaling fails to prevent collapse, mixing real data with quality-aware filtering effectively preserves diversity. Ultimately, our results suggest that without policy-mandated human oversight, the deployment of generative AI threatens to degrade the very healthcare data ecosystems it relies upon.