Development of mixed histiocytosis (Langerhans cell histiocytosis (LCH))/Erdheim–Chester disease (ECD)) after treatment in patients with an initial skull LCH lesion has not been well recognized. An elderly woman initially developed LCH at the left temporal bone, preceded by polyuria and polydipsia five years earlier; the lesion was surgically removed. Two years thereafter, she experienced her first LCH relapse with a right parietal skull lesion, in which a BRAF V600E mutation was confirmed, and chemotherapy was initiated. After a second LCH relapse involving the left parietal bone, the patient presented with a third relapse at the L2 vertebra. This lesion was pathologically diagnosed as mixed histiocytosis (LCH/ECD), resulting in refractoriness to conventional chemotherapy, and was successfully treated with targeted therapy using BRAF and MEK inhibitors. Spinal mixed histiocytosis (LCH/ECD) may develop following relapses of skull LCH after chemotherapy, for which targeted therapy could be effective.

Normalization can standardize and improve machine learning (ML) performance on omics data. However, it is unclear whether normalization is associated with overfitting (i.e., worse cross-dataset performance than intra-dataset performance). Therefore, we aimed to examine associations of normalization and regularization with overfitting of ML on omics data.

Using three paired transcriptomic and clinical datasets (lung adenocarcinoma: the Cancer Genome Atlas (TCGA)/Oncology Singapore; melanoma: TCGA/Dana-Farber Cancer Institute; glioblastoma: TCGA/Clinical Proteomic Tumor Analysis Consortium), we applied ANOVA-based gene selection methods, six normalization methods, and six ML models to classify cancer patients’ deaths. Balanced accuracy (BA) and area under the curve (AUC) in intra- and cross-dataset settings were compared using inferential analyses.

Normalization consistently improved intra-dataset performance (median BA/AUC changes: 0.035–0.214/0.115–0.279) on all data, particularly with Z_Raw, but decreased or slightly increased cross-dataset performance (median BA/AUC changes: −0.029–0.079/0.029–0.064). Least Absolute Shrinkage and Selection Operator (LASSO) model without normalization consistently outperformed most of the ML models in cross-dataset testing across cancer types. ML models on all and molecular-alone data showed similar best performances.

Normalization increases ML’s intra-dataset performance and overfitting in three paired cancer transcriptomic and clinical datasets. Regularized models such as LASSO appear to mitigate overfitting and achieve robust cross-dataset performance. Therefore, cross-dataset evaluation and regularized models are recommended to assess and reduce overfitting, while normalization should be used cautiously. Adding clinical data seems to have little impact on ML models’ performance. However, future work on other diseases and datasets is warranted.

Hepatocellular carcinoma (HCC) remains one of the most fatal cancers, primarily due to late diagnosis and the lack of effective early biomarkers. Recent advances in multi-omics and liquid biopsy technologies hold promise for improving early detection, prognostication, and monitoring of HCC. Understanding the immune landscape of HCC through genetic and epigenetic signatures is essential for identifying therapeutic targets and improving immunotherapy outcomes. This review aims to present current findings on immune-related biomarkers, multi-omics strategies, and biomarker validation in HCC. It also aims to evaluate the role of liquid biopsy and gene signatures in predicting treatment responses, with a specific focus on their applications in immunotherapy. The goal is to provide a comprehensive framework for integrating these emerging tools into clinical practice. The integration of multi-omics approaches has led to the identification of robust gene signatures that predict HCC prognosis and response to immune checkpoint inhibitors. Liquid biopsy technologies, including circulating tumor DNA, provide non-invasive alternatives for monitoring tumor evolution and therapeutic responses. Despite promising results, challenges remain in clinical validation, particularly in cross-platform reproducibility and the interpretation of complex multi-omics data. While genetic biomarkers are rapidly advancing, their clinical application in personalized medicine remains hindered by technical and ethical challenges, such as data privacy, informed consent, and method standardization. The integration of multi-omics data and liquid biopsies offers a promising path toward real-time, personalized treatment and the development of universal prognostic signatures for HCC. However, successful clinical adoption depends on cross-disciplinary collaboration to standardize data protocols and overcome challenges regarding accuracy, reproducibility, and patient privacy.

Synucleinopathies, including Parkinson’s disease (PD), dementia with Lewy bodies, and multiple system atrophy (MSA), are a group of neurodegenerative diseases characterized by the oligomerization of α-synuclein protein in neurons or glial cells. Various splicing isoforms of α-synuclein have been described, each with different aggregation properties. The α-synuclein gene (SNCA) has been identified as a highly significant genetic risk locus associated with various synucleinopathies across populations. This study aimed to assess the association of SNCA genetic variants with MSA and the levels of SNCA transcripts in peripheral blood mononuclear cells (PBMCs) from MSA and PD patients.

In this retrospective case–control study, 96 MSA patients, 1086 PD patients, and 485 healthy volunteers were included. PCR followed by restriction endonuclease analysis was used to detect four SNCA single-nucleotide polymorphisms (rs356219, rs3756063, rs11931074, and rs356168) in these individuals. In addition, RT-qPCR was performed to detect the levels of α-synuclein transcripts (SNCA mRNA isoforms -140, -126, and -112) in PBMCs of 24 MSA patients (including parkinsonian (MSA-P) and cerebellar (MSA-C) variants), 31 PD patients, and 32 healthy volunteers.

The frequency of the ‘T’ allele (of rs11931074) was significantly higher in MSA patients than in the healthy controls. The level of SNCA-140 mRNA was significantly decreased in MSA and PD patients compared with the controls, while the level of SNCA-112 mRNA was significantly increased in MSA-P patients than in PD patients and the controls. SNCA-112 mRNA/SNCA-140 mRNA and SNCA-112 mRNA/SNCA-126 mRNA ratios were significantly increased in MSA patients than in the controls.

The SNCA rs11931074 polymorphism is associated with MSA. There is a pronounced alteration in the expression of SNCA transcripts in PBMCs of MSA and PD patients.

Immunothrombosis, the interplay between immune activation and coagulation, contributes to disease progression in inflammatory disorders. Its role in hepatitis B virus–related acute-on-chronic liver failure (HBV-ACLF) and the involvement of neutrophil extracellular traps (NETs) remain unclear. This study aimed to elucidate NETs-mediated immunothrombosis in HBV-ACLF.

Liver single-cell RNA sequencing data from HBV-ACLF patients and healthy controls were analyzed to define immune and endothelial transcriptional profiles. A cohort of 46 HBV-ACLF patients, 20 chronic hepatitis B patients, and 20 healthy controls was assessed for circulating NETs, endothelial injury markers, and coagulation parameters. Histopathology and in vitro assays examined NETs distribution and endothelial interactions.

NETs were markedly elevated in HBV-ACLF and correlated with endothelial injury markers (syndecan-1, von Willebrand factor, soluble thrombomodulin), coagulopathy, and prognostic scores. Histology revealed NETs colocalization with endothelial cells and platelets within hepatic microthrombi. NETs from patient neutrophils impaired endothelial integrity and enhanced procoagulant activity in vitro. Mechanistically, toll-like receptor 2 (TLR2) and complement component 5a receptor 1 (C5aR1) signaling were involved in NETs formation, and their pharmacological inhibition reduced NETs generation.

NETs are associated with endothelial injury and immunothrombosis in HBV-ACLF. Mechanistic analyses suggest a role for TLR2 and C5aR1 pathways in NETs formation, indicating potential targets for future therapeutic investigation.

Chikungunya fever, caused by the Chikungunya virus (CHIKV), has re-emerged as a significant global health concern in recent decades. A notable event was the largest-ever local outbreak in China in 2025, marking a critical juncture in its epidemiology. Although conventional treatment remains predominantly supportive, the integration of traditional Chinese medicine (TCM) offers promising complementary strategies for alleviating both acute symptoms and chronic polyarthralgia. This narrative review aims to consolidate current knowledge on the etiology, pathogenesis, clinical manifestations, and management of Chikungunya fever, with a particular focus on the evidence-based application of TCM. By integrating molecular virology with clinical and epidemiological insights, this review offers a comprehensive perspective on the challenges posed by CHIKV and underscores the strategic imperatives essential for its future management. In conclusion, addressing the expanding threat of CHIKV necessitates a multi-pronged public health strategy that integrates standard clinical and preventive measures with evidence-based TCM therapies, highlighting the urgent need for rigorous clinical trials to globally validate these integrative treatments.

The immunosuppressive tumor microenvironment (TME) limits immunotherapy efficacy in intrahepatic cholangiocarcinoma (ICC). Understanding the molecular drivers of this TME is essential for developing new therapies. This study aimed to identify novel oncogenes that modulate the immune landscape of ICC using a multi-omics approach.

We integrated transcriptomic and proteomic data from our ICC cohorts with public datasets (TCGA-CHOL, GSE107943, OEP002768) to identify genes co-upregulated with PD-L1 (CD274). Single-cell RNA sequencing (scRNA-seq) was used to analyze cell-type-specific expression and intercellular communication. Clinical significance was validated through tissue microarrays and multiplex immunofluorescence in an independent ICC cohort.

Multi-omics screening identified TACC3 as a key candidate in ICC. Elevated TACC3 expression in ICC tissues correlated with poor prognosis and promoted tumor cell proliferation and migration. TACC3 activated the STAT3 pathway, increasing PD-L1 transcription. scRNA-seq showed TACC3/PD-L1 interaction in malignant epithelial cells, with PD-L1 co-expressed with FOXP3 in regulatory T cells (Tregs). Cell–cell communication analysis predicted strong interactions between malignant cells and Tregs. TACC3 knockdown reduced PD-L1 expression and inhibited STAT3 and AKT phosphorylation. Clinical validation confirmed co-expression of TACC3, PD-L1, and FOXP3, with high TACC3 levels linked to worse clinicopathological features and shorter progression-free survival.

Our study defines a TACC3-STAT3-PD-L1 axis driving immunosuppression in ICC. TACC3 fosters an immunosuppressive TME by upregulating PD-L1 and is associated with a Treg-rich contexture, suggesting that TACC3 may serve as a potential therapeutic target to overcome ICC immunosuppression.

Metabolic dysfunction-associated fatty liver disease (MAFLD) represents a predominant cause of chronic liver disease, underscoring the demand for accessible, non-invasive diagnostic tools. Tongue diagnosis in Traditional Chinese Medicine provides a distinctive perspective on systemic health, though it remains largely subjective. This study aimed to develop an interpretable multimodal deep learning model for MAFLD screening by integrating quantitative tongue image features with routine clinical data.

From 904 screened candidates, 477 subjects (157 healthy, 320 MAFLD) were included and randomly allocated to training, validation, and test sets in an 8:1:1 ratio. All participants underwent standardized tongue imaging (International Commission on Illumination L*a*b color features) and comprehensive clinical evaluation. We constructed a dual-stream deep learning model, combining a ConvNeXt-Tiny network for tongue images and a multilayer perceptron for clinical variables. Feature fusion was achieved via a Dynamic Affine Feature Transformation module, and the model was trained using weighted cross-entropy loss.

MAFLD patients showed significant metabolic abnormalities compared to healthy controls. A progressive decrease in tongue yellowness (b* value) was observed with advancing fibrosis. On an independent test set (n = 48), the multimodal model achieved 97.92% accuracy, Quadratic Weighted Kappa of 0.9538, and 96.88% sensitivity, and 100% specificity, outperforming single-modality and serological models. Interpretability analyses confirmed the model’s focus on clinically relevant tongue regions and key metabolic drivers.

We developed an accurate and interpretable multimodal model that synergizes tongue image features with metabolic indicators for MAFLD screening. This approach presents a promising, low-cost tool potentially well-suited for resource-limited settings.

Artificial intelligence (AI) is increasingly reshaping diagnostic pathology, with breast pathology representing one of the most advanced and clinically impactful areas of adoption. Despite rapid progress, many practicing pathologists remain unfamiliar with core AI concepts and their practical implications. This review provides a concise and accessible overview of AI in breast pathology, focusing on foundational principles, current clinical applications, and future directions.

Pertinent literature was reviewed. Personal experiences were also summarized and incorporated.

Key AI concepts, including algorithms, models, architectures, machine learning, deep learning, neural networks, and multimodal and foundational models, are introduced to establish a common framework. Important distinctions among generative, black-box, and explainable AI are highlighted, emphasizing the need for transparency and interpretability in clinical settings. The evolution of AI in breast pathology is reviewed, from early rule-based computer-assisted diagnostic systems to modern deep learning approaches leveraging large-scale whole-slide imaging datasets. Current applications span multiple domains, including detection of lymph node metastases, Nottingham grading, classification of benign and malignant lesions, and automated quantification of critical biomarkers. AI-based approaches to prognosis, risk stratification, prediction of treatment response, and analysis of the tumor microenvironment are also discussed. Finally, the review addresses challenges associated with real-world implementation, including data quality, bias, regulatory considerations, cost, infrastructure, and workflow integration.

As AI continues to evolve toward large-scale, multimodal, and explainable models, it is expected to function as an augmentative tool rather than a replacement for pathologists, supporting diagnostic accuracy, standardization, and personalized management in breast cancer care.