Accurate classification and grading of lymphoma subtypes are essential for treatment planning. Traditional diagnostic methods face challenges of subjectivity and inefficiency, highlighting the need for automated solutions based on deep learning techniques.

This study aimed to investigate the application of deep learning technology, specifically the U-Net model, in classifying and grading lymphoma subtypes to enhance diagnostic precision and efficiency.

In this study, the U-Net model was used as the primary tool for image segmentation integrated with attention mechanisms and residual networks for feature extraction and classification. A total of 620 high-quality histopathological images representing 3 major lymphoma subtypes were collected from The Cancer Genome Atlas and the Cancer Imaging Archive. All images underwent standardized preprocessing, including Gaussian filtering for noise reduction, histogram equalization, and normalization. Data augmentation techniques such as rotation, flipping, and scaling were applied to improve the model's generalization capability. The dataset was divided into training (70%), validation (15%), and test (15%) subsets. Five-fold cross-validation was used to assess model robustness. Performance was benchmarked against mainstream convolutional neural network architectures, including fully convolutional network, SegNet, and DeepLabv3+.

The U-Net model achieved high segmentation accuracy, effectively delineating lesion regions and improving the quality of input for classification and grading. The incorporation of attention mechanisms further improved the model's ability to extract key features, whereas the residual structure of the residual network enhanced classification accuracy for complex images. In the test set (N=1250), the proposed fusion model achieved an accuracy of 92% (1150/1250), a sensitivity of 91.04% (1138/1250), a specificity of 89.04% (1113/1250), and an F1-score of 90% (1125/1250) for the classification of the 3 lymphoma subtypes, with an area under the receiver operating characteristic curve of 0.95 (95% CI 0.93-0.97). The high sensitivity and specificity of the model indicate strong clinical applicability, particularly as an assistive diagnostic tool.

Deep learning techniques based on the U-Net architecture offer considerable advantages in the automated classification and grading of lymphoma subtypes. The proposed model significantly improved diagnostic accuracy and accelerated pathological evaluation, providing efficient and precise support for clinical decision-making. Future work may focus on enhancing model robustness through integration with advanced algorithms and validating performance across multicenter clinical datasets. The model also holds promise for deployment in digital pathology platforms and artificial intelligence-assisted diagnostic workflows, improving screening efficiency and promoting consistency in pathological classification.

We analyzed the clinical efficacy and safety of conversion therapy for unresectable hepatocellular carcinoma (uHCC) using transarterial chemoembolization (TACE) plus tislelizumab and characterized its effects on the tumor and immune landscape.

Fifty-seven patients with uHCC undergoing TACE plus tislelizumab from March 2020 to October 2024 were potentially eligible, and thirty-two patients finally enrolled. The efficacy endpoints included successful conversion rate, objective response rate (ORR), overall survival (OS), and progression-free survival (PFS). Tumor response was assessed using modified Response Evaluation Criteria in Solid Tumors (mRECIST). Treatment-related adverse events (TRAEs) were recorded according to CTCAE v5.0. Hematoxylin and eosin (HE) staining was used to evaluate tumor necrosis and lymphocyte infiltration. Immunohistochemistry (IHC) was performed to differentiate tumor immune infiltrates via a set of markers.

The best overall responses were 9.4% CR, 46.9% PR, 37.5% SD, and 6.3% PD, the ORR was 56.3%. Better ORR was shown in patients with AFP ≥ 400 ng/mL or tumor number < 3. The median PFS and OS was 13.9 (95% CI 2.6-25.2) and 29.2 (95% CI 13.7-44.7) months, respectively. Thirty-two patients (100%) experienced TRAEs of any grade, eight patients (25%) experienced grade 3/4 TRAEs. Fifteen patients (46.9%) with uHCC successfully converted. Notably, HE staining revealed extensive tumor necrosis and massive infiltration of lymphocytes in HCC and at the tumor-non-tumor interface in converted specimens. Further characterization by IHC revealed increased infiltration of CD8 + T and Th1 cells in the tumor of converted patients.

TACE plus tislelizumab may be a potent and safe conversion regimen for uHCC due to its ability to generate profound tumor necrosis and anti-tumor immune response.

Triple-negative breast cancer (TNBC) is highly malignant with rising incidence and mortality. Solute carrier family 7 member 5 (SLC7A5) is an amino acid transporter, and its mechanism in TNBC is still unclear. The public databases (TIMER2.0, TCGA, GEPIA, Kaplan-Meier Plotter, linkedomics, RBPmap, and RBPDB) were used to analyze the expression and correlation of genes and prognosis correlation. Gene and protein expression was detected by quantitative reverse transcription polymerase chain reaction (qRT-PCR) and western blot. 3-(4, 5)-Dimethylthiahiazo (-z-y1)-3, 5-di-phenytetrazoliumromide (MTT), terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick end labeling (TUNEL) staining, flow cytometry, transwell, sphere/tube formation, and metabolic kits were used to assess cell viability, apoptosis, invasion, stemness, angiogenesis, and glutamine (Gln) metabolism. The binding of forkhead box M1 (FOXM1) to the SLC7A5 promoter was determined by dual-luciferase reporter assay/chromatin immunoprecipitation (ChIP). RNA binding protein immunoprecipitation (RIP), RNA pull-down, and actinomycin D (Act D) experiments evaluated fused in sarcoma (FUS)-SLC7A5 interaction. In vivo, the mouse xenografts were established to validate the effects of FOXM1/SLC7A5 axis. Hematoxylin-eosin (HE) and immunohistochemistry (IHC) assays were used for histological morphology analysis and the protein expression of Ki67 and SLC7A5. SLC7A5 was up-regulated in TNBC and correlated with poor prognosis. Its knockdown repressed TNBC cell viability, invasion, stemness, angiogenesis, and Gln metabolism (as evidenced by reduced Gln, α-ketoglutarate (α-KG), and adenosine 5'-triphosphate (ATP) levles). FOXM1 transcriptionally activated SLC7A5; SLC7A5 overexpression reversed the suppressive effects of FOXM1 knockdown on TNBC malignancy and metabolism. Additionally, FUS bound to and stabilized SLC7A5 mRNA. In vivo, SLC7A5 counteracted the FOXM1 knockdown-mediated inhibition of tumor growth and the reduction in SLC7A5 and Ki67 protein expression. In conclusion, SLC7A5 promotes TNBC malignancy and metabolism. Its expression is transcriptionally driven by FOXM1 and post-transcriptionally stabilized by FUS. Targeting the FOXM1/SLC7A5 axis presents a novel therapeutic strategy for TNBC.

The development of neoantigen-based cancer vaccines has emerged as a groundbreaking approach in the field of personalized oncology. Neoantigens, originating from tumor-specific somatic mutations, possess considerable immunogenic potential and are absent in normal tissues, making them ideal candidates for eliciting targeted and enduring anti-tumor immune responses. Progress in next-generation sequencing, immunopeptidomics, and computational epitope prediction, has accelerated the identification and prioritization of patient-specific neoantigens, thereby facilitating the development of diverse vaccine platforms, including peptide, mRNA, DNA, and dendritic cell-based formulations. Initial clinical trials have demonstrated the safety, practicality, and immunogenicity of neoantigen vaccines in various cancers, producing promising therapeutic responses, particularly when combined with immune checkpoint inhibitors. Despite these advancements, substantial challenges-such as tumor heterogeneity, the accuracy of neoantigen prediction, immune evasion mechanisms, and manufacturing complexities-continue in impede widespread clinical application. This study provides a comprehensive analysis of neoantigen biology, advanced detection technologies, and delivery platforms, while meticulously assessing clinical outcomes, combinatorial strategies, and existing limitations. It also highlights new opportunities, such as the use of artificial intelligence and the mass production of vaccines. Neoantigen-based vaccines represent a significant breakthrough in cancer immunotherapy, offering highly individualized, tumor-targeted treatment strategies that could improve long-term patient survival.

Pseudokinases, a subclass of the human kinome, play critical roles in cellular regulation despite their lack of enzymatic activity. Among these, the Receptor Tyrosine Kinase-Like Orphan Receptor 1 (ROR1) has emerged as a pivotal regulator in cancer biology. ROR1, initially identified for its role in embryonic development, is aberrantly expressed in various malignancies, including hematologic cancers and solid tumors, while being largely absent in normal adult tissues. Its unique expression profile, coupled with its role in tumor survival, metastasis, therapy resistance, and stemness, makes ROR1 an attractive therapeutic target. This review provides an in-depth analysis of the structural and functional attributes of ROR1, its interaction with oncogenic signaling pathways, and its implications in tumor progression. We also explore current therapeutic strategies, including monoclonal antibodies, antibody-drug conjugates (ADCs), chimeric antigen receptor (CAR) T-cell therapy, and small-molecule inhibitors, highlighting their clinical potential and limitations. Understanding the mechanistic underpinnings of ROR1-driven oncogenesis and overcoming therapeutic challenges will be crucial for developing effective anti-cancer strategies targeting this pseudokinase.

Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer characterized by high rates of tumor protein 53 (TP53) mutation and with limited targeted therapies. Despite being clinically advantageous, direct targeting of mutant TP53 has been challenging. Therefore, we hypothesized that p53-mutant TNBC cells rely upon other potentially targetable survival pathways.

In vitro and in silico screens were used to identify drugs that induced preferential death in TP53-mutant cells. The effect of the ferroptosis inducer ML-162 was tested both in vitro and in vivo and the mechanism of cell death following ML-162 treatment or GPX4 knockout was determined.

High-throughput drug screening demonstrated that TP53-mutant TNBCs are highly sensitive to peroxidase, cell cycle, cell division, and proteasome inhibitors. We further characterized the effect of the ferroptosis inducer ML-162 and demonstrated that ML-162 induces preferential ferroptosis in TP53-mutant TNBC cells. Treatment of TP53-mutant xenografts with ML-162 suppressed tumor growth and increased lipid peroxidation in vivo. Testing ferroptosis inducers demonstrated TP53-missense mutant, and not TP53-null or wild-type cells, were more sensitive to ferroptosis, and expression of mutant TP53 genes in p53-null cells sensitized cells to ML-162 treatment.

This study demonstrates that TP53-mutant TNBC cells have unique survival pathways that can be effectively targeted. Our results illustrate the intrinsic vulnerability of TP53-mutant TNBCs to ferroptosis and highlight GPX4 as a potential target for the precision treatment of TP53-mutant TNBC.

Advanced technical developments of small-caliber disposable endoscopes as well as EUS imaging and EUS-assisted procedures have changed the management of biliary and pancreatic diseases and expanded the spectrum of diagnostics and endoscopic interventions. Cholangioscopy enables the direct visualization of the biliary system. Endoscopic and endosonographic (EUS) procedures nowadays provide versatile and often minimally invasive approaches for assessment of the pancreatic duct system. The digital single-operator cholangioscopy (dSOC) and digital single-operator pancreatoscopy (dSOP) are increasingly becoming an integral component of the treatment of various diseases and disorders of both duct systems. This article familiarizes the reader with the various diseases and disorders of the biliary duct and pancreatic duct system and presents the most important indications, techniques of the diagnostics and therapeutic interventions as well as their clinical value.

Hepatocellular carcinoma (HCC) is a highly prevalent and lethal malignancy with complex pathogenesis and limited treatment options. Exosomes play crucial roles in HCC development and progression. This study aimed to comprehensively investigate the differential expression and functional roles of lipocalin-2 (LCN2) in HCC exosomes. Bioinformatics analysis of the GSE199509 dataset identified 4518 differential expression genes (DEGs) between HCC patient-derived exosomes and normal controls, with LCN2 being highly expressed in HCC exosomes. In vitro experiments showed that LCN2 was significantly enriched in exosomes from HCC cell lines (SMMC-7721 and HepG2). HCC-derived exosomes induced M2 macrophage polarization, which promoted HCC cell migration, invasion, and proliferation. Knockdown of LCN2 attenuated HCC exosome-induced M2 macrophage polarization and reduced the pro-tumor effects on HCC cells. Mechanistically, LCN2 inhibited Nedd4-1-mediated ubiquitination and degradation of SR-BI in a Nedd4-1-dependent manner. SR-BI overexpression reversed the effects of LCN2 knockdown on macrophage polarization and HCC cell proliferation. In vivo mouse models demonstrated that LCN2 knockdown inhibited tumor growth in wild-type mice but not in SR-BI knockout mice. These findings suggest that LCN2 plays a pivotal role in HCC progression through modulating macrophage polarization and interacting with SR-BI. Targeting the LCN2-SR-BI axis may represent a novel therapeutic strategy for HCC. Additionally, exosomal LCN2 could be a potential biomarker for HCC diagnosis and prognosis. However, further pre-clinical and clinical studies are needed to validate these findings.

Chimeric antigen receptor (CAR) technology has revolutionized cancer therapy, yet its full potential remains untapped within the innate immune system. Beyond CAR-T cells, a growing cadre of MHC-independent effectors, including NK cells, macrophages, γδ T cells and the emerging innate-like T cells such as invariant NKT (iNKT) and mucosal-associated invariant T (MAIT) cells, offer complementary mechanisms for tumor recognition and elimination. These platforms combine facile, off-the-shelf manufacture from healthy donors with low graft-versus-host disease risk and a reduced propensity for severe cytokine release syndromes. Mechanistically, they span missing-self and antibody-dependent cytotoxicity (NK), phagocytosis and cross-presentation (macrophages), stress-ligand recognition (γδ T cells), and rapid, tissue-tropic, TCR-mediated responses to conserved lipid and metabolite antigens (iNKT via CD1d; MAIT via MR1). CAR engineering of these cells leverages their innate rapidity, innate/adaptive cross-talk, and distinctive homing to confront heterogeneous and immune-evasive tumors. Here, we synthesize recent advances in cell design, dual/split CARs, switchable control systems, armored payloads and synthetic-biology circuits, and evaluate translational progress, manufacturing bottlenecks, and regulatory considerations. We argue that integrating innate and innate-like programs with precision CAR architectures will yield a new generation of universal, resilient cellular therapeutics with broadened antigen reach, improved safety profiles, and enhanced capacity to overcome the suppressive tumor microenvironment.