Colon adenocarcinoma (COAD) is a heterogeneous malignancy whose molecular complexity limits effective therapy. Existing transcriptome-based classifications capture only part of this diversity. To refine COAD stratification, we integrated genomic, epigenomic, and transcriptomic data from 297 The Cancer Genome Atlas patients. Ten complementary clustering algorithms were combined through a consensus ensemble framework to ensure robust and unbiased subtype discovery. The resulting molecular subtypes were characterized by genomic alterations, signaling pathways, tumor microenvironment features, and predicted therapeutic responses. As a result, four reproducible molecular subtypes (CS1-CS4) were identified. CS1 displayed enrichment of extracellular matrix organization and epithelial-mesenchymal transition signatures, suggesting invasive potential. CS2 exhibited transcriptional similarity to PD-1 responders, indicating potential benefit from immune checkpoint blockade. CS3 represented a mutation-driven subtype with frequent APC, TP53, and KRAS alterations and extensive copy number gains. CS4 showed the highest immune infiltration, elevated tumor mutational burden, and enhanced sensitivity to 5-fluorouracil and cetuximab. Validation across four independent cohorts confirmed the reproducibility of these subtypes. This integrative multi-omics framework refines the molecular taxonomy of COAD, revealing immunologically active and therapeutically distinct subgroups. The classification not only bridges genomic, epigenomic, and transcriptomic regulation but also provides a practical roadmap for precision oncology by linking molecular features to potential treatment strategies.

Background: This study aimed to explore the functions and potential mechanisms of PIWI-interacting RNA–related genes (piRPGs) in bladder cancer (BC) development and to identify potential prognostic genes. Methods: This study used differential analysis and machine learning techniques to identify the differentially expressed piRPGs in BC. Consensus clustering was performed on The Cancer Genome Atlas–Urothelial Bladder Carcinoma dataset, and univariate and multivariate analyses were conducted to construct a BC prognostic model consisting of 3 piRPGs. Kaplan-Meier survival curves were used for survival analysis. Quantitative polymerase chain reaction was performed to validate the expression levels of piRPGs in BC cells and tissues. The functional roles and potential mechanisms of piRPGs in BC were investigated via single-cell sequencing and differentially methylated position sequencing. The DSigDB and CellMiner databases were used to screen for small-molecule drugs associated with piRPGs. Results: This study identified 6 piRPGs that were significantly associated with BC: MAPK13, INHBA, LAMB2, DDX3X, TARBP2, and CDK2. A prognostic model comprising MAPK13, INHBA, and LAMB2 was constructed using consensus clustering technology. Kaplan-Meier curves demonstrated significantly prolonged survival in cluster 2 compared with cluster 1 ( p < 0.01), validating the effectiveness of the prognostic model. Single-cell sequencing confirmed that MAPK13 expression was significantly upregulated in bladder tissues ( p < 0.001). Methylation site sequencing and methylation-specific polymerase chain reaction revealed significantly decreased methylation levels of INHBA and MAPK13 in BC tissues, which were inversely correlated with their expression levels. Conclusions: This study effectively developed a 3-gene prognostic signature comprising MAPK13, INHBA, and LAMB2 using consensus clustering and multifactorial logistic regression. In addition, the functional roles and intrinsic mechanisms of piRPGs in bladder carcinogenesis were comprehensively explored using single-cell sequencing, methylation sequencing, and functional enrichment analysis.

Introduction Ulcerative colitis (UC) is a lifelong, chronic inflammatory disorder, characterized by recurrent and diffuse inflammation of the rectal and colonic mucosa. Increasing evidence suggests that impaired mitophagy contributes to immune dysregulation and epithelial injury in UC. However, the mitophagy-related molecular landscape and its therapeutic potential remain largely unexplored. Methods Mitophagy-related genes (MRGs) were intersected with differentially expressed genes to identify UC-associated MRGs. Functional enrichment, immune infiltration, and consensus clustering analyses were performed to characterize molecular subtypes. Three machine learning methods were employed to identify diagnostic models. Candidate therapeutic agents were identified by the CMap database. Results A total of 35 UC-associated MRGs were identified, enriched in cell activation, fatty acid metabolism, and the PPAR signaling pathway, revealing strong immunometabolic coupling in UC. Consensus clustering stratified UC patients into two subtypes: a metabolism-dominant subtype (C1) and an inflammation-activated subtype (C2). Three hub genes—CD55, CPT1A, and SLC16A1—were screened and validated as robust diagnostic markers. Drug prediction and molecular docking revealed strong binding between galunisertib and CD55, which was further validated by molecular dynamics simulations. In vitro , galunisertib significantly suppressed inflammatory cytokine release in LPS-induced UC cell models. Discussion This study delineated the mitophagy-related molecular signatures of UC and identified CD55, CPT1A, and SLC16A1 as key biomarkers linking mitochondrial dysfunction, metabolic reprogramming, and immune activation. Furthermore, galunisertib was proposed as a potential therapeutic agent, providing a theoretical basis for UC therapy.

Development of a lactylation-related molecular classification and machine learning-based gene signature to predict survival, response to immunotherapy for ovarian cancer.

Drug resistance, characterized by high heterogeneity and complex mechanisms, poses a significant challenge in cancer treatment. Stratifying resistant tumors into biologically and clinically meaningful subgroups can improve prognostic evaluation and help guide treatment decisions. However, the DNA methylation-based subtypes of resistant tumors have not yet been comprehensively characterized. DNA methylation profiles from resistant tumors were retrieved from public database including TCGA and GEO. For each tumor type resistant to a specific treatment drug, consensus clustering based on the most variable methylated probes was conducted to identify the DNA methylation subtypes of resistant tumors. For low-grade glioma (LGG) resistant to Temozolomide, consensus clustering of highly variable CpGs identified two subtypes: cancer resistance CpG island methylator phenotype-positive (CR_CIMP+) and -negative (CR_CIMP-). The CR_CIMP- subtype associates with poorer prognosis, reduced drug response, and more advanced histology, exhibiting higher tumor mutation burden and greater activity in drug resistance-related pathways, such as PI3K/AKT/mTOR signaling. CR_CIMP subtypes with distinct clinical or molecular features were also identified in pancreatic adenocarcinoma and bladder urothelial carcinoma resistant to Gemcitabine, as well as in non-small cell lung cancer resistant to anti-PD1/PD-L1 immunotherapy. Based on predicted drug responses, the study screens candidate drugs for each CR_CIMP subtype. Finally, a random forest model is proposed to predict CR_CIMP subtypes in LGG patients resistant to Temozolomide. This study uncovers DNA methylation subtypes within resistant tumors, enabling more precise stratification to inform prognosis and therapy selection.

Degenerative lumbar spondylolisthesis (DLS) is the commonest reason for decompression and fusion in the elderly. Current evidence questions the necessity of accompanying fusion in all DLS because of significant clinical and radiographic heterogeneity. This heterogeneity creates substantial variability in surgeon reported decision-making and identification of the ideal patient for decompression vs. decompression and fusion. Our goal was to identify distinct DLS groups, aiming to inform clinical decision-making for decompression vs. decompression and fusion. Consensus cluster analysis was used based on demographic, clinical, and radiological characteristics of patients enrolled in the CSORN (Canadian Spine Outcomes and Research Network) prospective DLS study from 2015 to 2022 at eight Canadian sites. We selected clinically relevant variables to guide feature selection and evaluated the optimal number of clusters. Euclidean distance metric was used to determine the cluster number and structure, with up to seven clusters considered and 50 sampling iterations. Key characteristics of each cluster were identified, and post-surgical outcomes were compared using the Oswestry Disability Index (ODI) with pairwise statistical tests. Consensus cluster analysis of 486 DLS patient (mean [SD] age, 55.4 [9.2] years, 180 men [37%]) identified four distinct clusters. Cluster 1 (n=175) included patients with moderate disability, low comorbidity, low depression, and well-aligned spine. Cluster 2 (n=85) comprised predominantly female patients with severe disability, high comorbidity, high depression, and poor spine alignment. Cluster 3 (n=91) mainly consisted of male patients with moderate disability and malaligned spine, while Cluster 4 (n=135) included elderly patients with severe disability and relatively well-aligned spine. Clusters 1 and 3 showed the most favorable ODI outcomes at 1- and 2-years follow-up. Cluster 2 had the highest rate of fusion surgery (84%) and the greatest ODI change at 1 year (DODI mean [SD] 27.97 (19.32)). Decompression alone provided sustained benefit in Cluster 1 and 3, while improvement regressed in Cluster 2 and 4 at two years. Unsupervised machine learning identified four clinically distinct clusters of DLS patients with varying outcome and responses to surgical interventions, highlighting the importance of stratified care strategies to optimize patient outcomes.

Breast cancer (BRCA) remains a leading cause of cancer-related mortality among women worldwide. The kinesin family member 5B (KIF5B) has been implicated in various cancers, yet its comprehensive role in BRCA prognosis, tumor microenvironment (TME), and therapeutic response remains poorly understood. We integrated single-cell RNA sequencing (scRNA-seq) data and bulk RNA-seq data from multiple datasets. Using hdWGCNA, we identified gene modules related to the unfolded protein response (UPR). Consensus clustering defined UPR-related molecular subtypes, and differential expression analysis revealed key prognostic genes. KIF5B was further evaluated for its prognostic significance, mutational landscape, immune infiltration associations, and potential as a therapeutic target. Malignant cell-specific KIF5B upregulation drives poor overall survival in pan-cancer cohorts, validating its hazardous molecular function. Functional enrichment analysis linked KIF5B to critical pathways, including immune response, JAK-STAT signaling, and epithelial-mesenchymal transition. Immune infiltration analysis revealed that high KIF5B expression was associated with reduced immune and ESTIMATE scores, but higher tumor purity. Drug sensitivity prediction identified several compounds with potential efficacy in high KIF5B-expressing patients, including Sapitinib and LCL161. Our multimodal analysis establishes KIF5B as a prognostic biomarker and potential therapeutic target in BRCA, with implications for understanding immune evasion and guiding precision treatment strategies.

Propionate exerts antiproliferative and immunomodulatory effects in tumors, but its role in the metabolism of lung adenocarcinoma (LUAD) remains underexplored. This study aimed to characterize molecular subtypes based on propionate metabolism-related genes (PMRGs) and assess their prognostic and immunological relevance in LUAD. Using transcriptomic data from The Cancer Genome Atlas (TCGA)–LUAD and validation from GSE30219, consensus clustering was performed to identify subtypes associated with propionate metabolism. Immune infiltration and tumor microenvironment characteristics were analyzed through established algorithms. Differentially expressed genes (DEGs) were identified, and a prognostic model was constructed using Cox and least absolute shrinkage and selection operator (LASSO) regression. Three molecular subtypes (low, medium, and high propionate metabolism) were identified, demonstrating significant differences in overall survival and immune microenvironment features. The high-propionate subtype was characterized by elevated immune and stromal scores, as well as increased M2 macrophage infiltration. A 7-gene prognostic signature was developed, with risk stratification revealing significant survival and drug sensitivity differences between high- and low-risk groups. Key prognostic genes, including SLC2A1, SLC16A1, IL1A, AHSG, and ALOX15, were validated through RT-qPCR. This study highlights the molecular heterogeneity of propionate metabolism in LUAD and proposes a prognostic signature that could inform immunotherapeutic stratification.

ObjectiveAdvanced prostate cancer (PCa) remains therapeutically challenging due to heterogeneous mechanisms of resistance to androgen receptor (AR)-targeting agents. While AR signaling persists in castration-resistant PCa (CRPC), emerging evidence suggests AR-independent survival pathways may contribute to therapeutic escape. This study integrates transcriptomic data and clinical profiling to dissect AR dependency and resistance mechanisms in PCa, aiming to identify subtype-specific vulnerabilities and therapeutic targets.MethodsWe performed CRISPR-Cas9 screens in AR-dependent (VCaP, LNCaP, 22Rv1) and AR-independent (DU145, PC-3, WPE1-NA22, P4E6, Shmac5) cell lines to identify core essential genes. RNA sequencing data from TCGA-PRAD, Changhai, and DKFZ cohorts were integrated to define molecular subtypes using consensus clustering. Spatial transcriptomics (ST) and single-cell RNA sequencing (scRNA-seq) were employed to validate gene expression patterns in primary tumors and metastatic samples. Temporal expression dynamics were analyzed using fuzzy clustering to identify resistance mediators, with a focus on MCL1. Drug sensitivity analysis revealed that AR-dependent cells were more sensitive to MCL1 inhibitor UMI-77, and MCL1 expression was higher in Enzalutamide-resistant cell lines. Functional validation via MCL1 knockdown confirmed its role in supporting the proliferation and inhibiting apoptosis of resistant cells.ResultsCRISPR screening identified 952 shared essential genes in prostate cancer, with 157 AR-high essential signature and 130 AR-low essential signature genes. AR-high essential signature genes enriched in cell cycle/polycomb pathways, while AR-low essential signature genes correlated with oxidative phosphorylation/mTOR signaling. Consensus clustering of TCGA-PRAD data revealed three molecular subtypes (Clusters 1–3); Cluster 3 showed worst prognosis (shorter PFI/OS) and advanced clinical features (higher T/N stage, Gleason grade). External validation confirmed Cluster 3’s aggressive phenotype and independent prognostic value (meta-cohort HR = 1.98, 95% CI: 1.19–3.27). Cluster 3 signature genes were upregulated in metastatic/CRPC tissues and spatially enriched in CRPC epithelium. Notably, Cluster 3 shared essential gene expression decreased after Enzalutamide treatment, whereas AR-high essential signature genes remained stable. MCL1 emerged as a key resistance driver, demonstrating persistent upregulation in Enzalutamide-resistant cells and CRPC models.ConclusionsThis study elucidates distinct AR dependency landscapes in PCa, revealing AR-independent survival pathways and a clinically actionable molecular subtype (Cluster 3) linked to therapy resistance. MCL1 emerges as a critical mediator of adaptive resistance, highlighting the need for combination therapies targeting both AR-driven and AR-independent programs to improve outcomes in advanced PCa.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12885-025-15357-5.

BackgroundAbdominal aortic aneurysm (AAA) refers to a lasting enlargement of the abdominal aorta. Senescence, a major risk factor of AAA, demonstrate positive connection with both the formation and rupture of aneurysms. Therefore, investigating the underlying pathogenic mechanisms of senescence in AAA and exploring relevant diagnostic and therapeutic targets is crucial.MethodsThree transcriptomic datasets related to AAA were obtained from the GEO database, and collection of genes associated with cellular senescence was obtained from MSigDB. Overlapping genes of differentially expressed genes (DEGs), module genes associated with AAA, and senescence-related gene sets were identified as senescence-related DEGs of AAA and subjected to further functional enrichment analysis. Distinct machine learning algorithms were subsequently utilized to screen for senescence-associated biomarkers and develop a diagnostic nomogram. In addition, the interaction between these biomarkers and immune components in the aneurysmal environment were revealed. Consensus clustering was subsequently applied to classify AAA into distinct subtypes. Finally, validation was performed using an AAA murine model.ResultsA total of 11 senescence-related DEGs in AAA were identified, which mainly involved with oxidative stress, inflammatory responses, and vascular smooth muscle cell activity. Following rigorous screening, IL6, ETS1, TDO2, and TBX2 were identified as diagnostic biomarkers for senescence-related DEGs of AAA. The nomogram constructed from these biomarkers demonstrated high discriminatory ability in the training cohort (AUC = 1), though this requires further validation in larger cohorts due to potential overfitting. Immune cell infiltration and single-cell analyses indicated that the expression of the diagnostic biomarkers is linked to various immune cell types. Consensus clustering identified two AAA subtypes, which exhibiting distinct expression patterns of senescence-related biomarkers. Finally, validation in an AAA murine model confirmed the expression changes of these senescence-related biomarkers in AAA.ConclusionThis study identified senescence-related biomarkers associated with AAA through transcriptomic public databases, revealing their potential functional mechanisms, relationships with immune cells, and associations with AAA subtypes. These results could offer novel candidate targets for both diagnostic and therapeutic strategies in AAA.

Abstract Background: Metastatic castration-resistant prostate cancer (mCRPC) is a heterogeneous disease and often comprise molecularly distinct subtypes that differ in prognosis and therapeutic response. Super-Enhancers (SEs) are large enhancer clusters that robustly drive expression of genes controlling developmental processes and cell identity. In this study, we developed an epigenetic-based framework to determine whether mCRPC harbors subtype-specific SE programs that may reactivate distinct developmental transcriptional programs and define molecularly subtypes of disease progression. Methods: We developed a computational workflow, Super-Enhancer Analysis for Lineages (SEAL), to map SE landscapes, classify tumor subtypes and identify subtype-specific SE-driven genes. Using ROSE analysis on H3K27ac ChIP-seq data from the LuCaP and MURAL PDX mCRPC series, we constructed an integrated SE atlas. Molecular subtypes were defined through consensus clustering of SE regions and further characterized by genomic, transcriptional, and clinical features. We integrated matched RNA sequencing data to identify top-ranked SE-driven transcription factors, which were further evaluated via loss- and gain-of-function studies as well as ChIP-seq and RNA-seq analyses to define their cistromes, transcriptomes, and interacting networks. Results: We identified five distinct SE programs in mCRPC: three AR-positive subtypes (AR-1, AR-2, AR-3) and two AR-independent subtypes (NEPC-like and DNPC-like). Among the three AR-driven subtypes, AR-1 and AR-2 displayed aggressive tumor features, enriched for cell cycle, EMT, and hypoxia pathways. Each subtype exhibited a distinct SE-driven transcriptional program. Notably, we identified TWIST1, HNF1A, and TBX10 as key SE-driven transcription factors in AR-1, AR-2, and AR-3 subtypes, respectively. Among these, HNF1A, a critical transcription factor involved in hepatic development and metabolic regulation, was exclusively expressed in the AR-2 subtype. HNF1A silencing significantly reduced proliferation in vitro and in vivo and markedly suppressed glycolytic activity. AR-2 tumors produced high levels of secreted albumin, a well-established HNF1A hepatic target, which was decreased upon HNF1A silencing. ChIP-seq and RNA-seq analyses revealed that HOXB13 co-occupied HNF1A-mediated enhancers and cooperatively regulated the hepatic transcriptional programs. Conversely, HNF1A overexpression in AR-3 cells induced the expression of hepatic markers, enhanced proliferation and migration, and recapitulated features of the AR-2 subtype, indicating that HNF1A acts as a driver of hepatic lineage reprogramming. Conclusions: Our study reveals a novel HOXB13-HNF1A transcriptional axis that governs a previously undefined hepatic reprogramming in a subset of AR-driven mCRPC tumors. These findings establish a SE-based molecular classification of mCRPC and highlight HNF1A as a key regulator of metabolic and lineage programs, with potential implications for precision therapy and biomarker development in advanced prostate cancer. Citation Format: Mingyu Liu, Songqi Zhang, Nolan D. Patten, Jared G. Lourie, Xiaolin Zi, Kai Zou, Shuai Gao, Kourosh Zarringhalam, Changmeng Cai. Super-enhancer landscape analysis reveals a HOXB13-HNF1A transcriptional axis driving hepatic reprogramming in castration-resistant prostate cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Prostate Cancer Research and Treatment; 2026 Jan 20-22; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86(2_Suppl):Abstract nr PR015.

Screening of kinase-related genes as diagnostic biomarkers and immune infiltration analysis in sepsis.

Lupus nephritis (LN), a severe manifestation of systemic lupus erythematosus, involves immune complex deposition, inflammation, and kidney damage. Recent studies indicate that pyroptosis, a pro-inflammatory cell death process, drives renal injury in LN. This study intended to identify key pyroptosis-related genes in LN using datasets from the GEO database, encompassing glomerular, tubulointerstitial, and whole kidney tissues from LN patients. Identified differentially expressed genes related to pyroptosis and created a predictive model using univariate and LASSO regression analysis. LN patients were classified into subtypes through consensus clustering. Immune microenvironment characteristics and hallmark pathways were further analyzed. Using the WGCNA, key gene modules and hub genes were recognized, followed by an analysis of their clinical relevance and distribution patterns using the Nephroseq database and scRNA-seq data. Cellular experiments were conducted to validate the findings. We identified 26 differentially expressed pyroptosis-related genes in LN glomeruli and created a 10-gene model with high diagnostic accuracy (AUC: 0.968 for tubulointerstitium, 0.990 for whole kidney). Consensus clustering divided LN into two subtypes: subtype1, characterized by inflammation and immune activation, and subtype2, characterized by cellular metabolism. WGCNA highlighted the grey60 module linked to subtype1, and identified GBP2 and EIF2AK2 as hub genes. Cellular experiments showed that GBP2 and EIF2AK2 were upregulated in LPS-stimulated macrophages and glomerular endothelial cells, and their siRNA-mediated knockdown triggered a decline in pyroptosis-related marker expression, implying their possible role as therapeutic targets for modulating pyroptosis in LN. In conclusion, GBP2 and EIF2AK2 show potential as candidate molecules for targeted therapy in LN.

Radiogenomic studies have mostly linked single-site radiomic features (RFs) to genomic alterations in locally-advanced lung cancer, limiting their applicability to patients with metastatic lung adenocarcinoma (MLUAD). Our aim was to evaluate associations between unsupervised CT-based radiomic clustering of single-site and multi-site features and oncogenic alterations (OAs) and response to treatment in MLUAD. Patients managed at our center (October 2016-January 2024) with pre-treatment CT scans and next-generation sequencing were retrospectively included. Reproducible RFs were extracted from all solid tumor lesions > 1 cm³ using an automated pipeline. Patient-level integration used the centroid of each patient's lesions in radiomic space, providing multi-site radiomics data. RFs from the largest and biopsied lesions were also isolated. Patients were clustered by unsupervised hierarchical consensus clustering using centroid-based (Cluster-C), largest lesion (Cluster-M), and biopsied lesion (Cluster-B) features. Uni- and multivariable associations with OAs (any OA, smoker-related [sOA], non-smoker-related [nsOA], or wild-type), overall response rate (ORR), and overall survival (OS) were investigated. Among 361 patients (median age 63.2 years; 41.3% women; 1721 segmented tumor lesions), 48.2% had sOA and 13% had nsOA. Cluster-M2 + M5 was enriched in KRAS (p = 0.048), MET (p = 0.046), and PI3KCA (p < 0.001) alterations. Cluster-M (especially Cluster-M2 + M5) independently predicted sOA (OR = 2.28, p = 0.006), and nsOA (OR = 5.49, p = 0.004). Cluster-M was linked to higher ORR (p = 0.026) and longer OS (p = 0.016). Baseline CT-based single- and multi-site radiomics capture patterns associated with key OAs in MLUAD, suggesting their potential role as a non-invasive adjunct to guide molecular testing and optimize treatment selection. Question In MLUAD, can single- and multi-site RFs from all measurable lesions enhance the detection of key OAs and outcome prediction beyond standard clinical-radiological assessment? Findings In 361 MLUAD patients, robust clustering using RFs from multiple tumor lesions per patient identified subgroups associated with key OAs, response to treatment, and survival. Clinical relevance Whatever the initial disease staging, radiomic clustering may serve as a non-invasive AI biomarker that complements molecular testing, helping identify actionable tumor profiles and stratify patients for treatment selection and prognostication in MLUAD.

BackgroundAlcohol-associated steatohepatitis (ASH) is a globally prevalent liver disease, with robust evidence implicating neutrophil extracellular traps (NETs) as a central pathological phenomenon driving inflammation and progression. However, the core genomic signatures that govern NETs and underlying molecular mechanisms within the ASH microenvironment remain poorly defined.MethodsBuilding on the prominent NETs formation phenomenon in ASH, we established a core pool of NETs-related hub genes through intersection of ASH-derived differentially expressed genes (DEGs), key WGCNA modules, and a curated NETs gene set. From this NET-focused pool, a consensus of three machine learning algorithms (LASSO, SVM, RF) distilled a final diagnostic signature, which was rigorously validated in training and external cohorts via ROC analysis and neural networks. Patient heterogeneity was then investigated using consensus clustering with this signature, followed by immune profiling and functional validation in human and mouse ASH models. Therapeutic potential was explored through drug database enrichment and molecular docking.ResultsA NETs-focused three-gene signature (FOS, MMP7, CXCL6) achieved exceptional diagnostic accuracy for ASH (AUC = 1.00 in training; 0.983 in validation). It stratified ASH into a Metabolic-dominant (C1) subtype and a Pro-inflammatory (C2) subtype, the latter exhibiting higher MMP7/CXCL6, lower FOS, and enriched cytotoxic infiltration. In vivo, FOS rose in acute injury but declined in chronic models and human ASH, whereas MMP7/CXCL6 remained elevated, suggesting a temporal shift from acute FOS-dominant response to sustained MMP7/CXCL6-mediated inflammation. Finally, drug-gene interaction analysis identified several potential therapeutic modulators, including N-acetylcysteine (NAC), with predicted high binding affinities to FOS and MMP7.ConclusionFOS, MMP7, and CXCL6 constitute a clinically actionable signature capturing the stage-specific dynamics of NETs-driven inflammation in ASH. Beyond its diagnostic and stratifying utility, this signature highlights potential therapeutic avenues for clinical intervention.

Liposarcomas (LPS) is a highly heterogeneous malignant soft tissue tumor. Tumor microenvironment immune traits critically affect cancer progression and treatment efficacy. However, immune-related biomarkers for prognostic assessment, reflecting tumor immune microenvironment features and with diagnostic potential, remain insufficiently explored in LPS. The RNA-seq data and clinical information of patients with liposarcoma were downloaded from the GEO and TCGA database. The "limma" package performed the differential expression genes (DEGs) analysis, and the weighted gene co-expression network analysis (WGCNA) method was used to identify the liposarcoma-related module. We performed the single sample gene set enrichment analysis (ssGSEA) to calculate the enrichment scores for 28 tumor-infiltrating lymphocyte (TIL) subpopulations based on previously established immune signatures. Then, the consensus clustering was conducted using the "ConsensusClusterPlus" package. After that, the lasso and multivariate Cox regression analysis was applied for the risk model construction. The ESTIMATE algorithm for immune infiltration, "clusterProfiler" for function enrichment, "survival" for prognostic difference and "timeROC" for receiver operator characteristic curve (ROC) analysis were performed. The wound healing and transwell assay were conducted. After differential expression analysis, 852 DEGs between liposarcoma and para-cancer samples were obtained, and the turquoise was the liposarcoma-related gene module through WGCNA analysis. Consensus clustering based on immune signatures stratified patients into immunity-high (H) and immunity-low (L) subtypes with significant survival differences. Integration of these findings led to a robust 2-gene prognostic signature (KNTC1 and PRC1) via LASSO and Cox regression. Both model genes exhibited outstanding diagnostic performance (AUC > 0.9). High RiskScore was significantly associated with aggressive pathological subtypes, metastasis, and an immunosuppressive microenvironment characterized by lower overall immune infiltration. Conversely, low-risk patients showed enhanced immune cell abundance and activity. In addition, functional validation confirmed that KNTC1 silencing significantly impaired LPS cell migration and invasion, underscoring its role in tumor progression. We constructed a robust two-gene prognostic model that effectively predicts survival, reflects tumor immune microenvironment heterogeneity, and distinguishes pathological subtypes in LPS. Our findings provided valuable insights for prognostic stratification and personalized treatment strategies.

This study aimed to develop and validate a Cuproptosis-Related Gene (CRG) signature for predicting Biochemical Recurrence-Free Survival (BCRFS) and characterizing the Tumor Immune Microenvironment (TIME) in Prostate Cancer (PCa). Transcriptomic and clinical data were collected from TCGA (n=405) and GEO (GSE70770, n=203). Consensus clustering based on 10 CRGs defined molecular subtypes. Differentially expressed genes between clusters were subjected to LASSO Cox regression to construct a prognostic signature in the TCGA cohort, followed by validation in GEO and combined cohorts. Quantitative real-time polymerase chain reaction (qRT-PCR) and Immunohistochemistry (IHC) were conducted for experimental validation. Two CRG-based subtypes were identified, characterized by distinct clinicopathological features, immune checkpoint expression, and BCRFS. A six-gene signature (CALML5, MMP11, UBE2C, ANPEP, TMEM59L, COMP) stratified patients into high- and low-risk groups with significantly different BCRFS (log-rank P<0.001). The model showed good predictive accuracy (AUCs 0.717-0.837 at 1 year, 0.728-0.771 at 3 years, 0.683-0.695 at 5 years) and remained independent of clinicopathological factors. High-risk patients exhibited elevated immune/stromal scores, altered immune infiltration, and higher immune checkpoint expression. qRT-PCR confirmed upregulation of CALML5, MMP11, UBE2C, and COMP in PCa cell lines, while IHC validated differential protein expression of all six genes between PCa and BPH tissues (all P<0.05). This six-gene CRG signature predicts BCRFS and reflects immune heterogeneity in PCa. Its integration into prognostic models may guide personalized management and inform immunotherapy strategies, warranting further validation in prospective clinical studies. This study initially identified two cuproptosis-related molecules based on the expression patterns of cuproptosis-related genes. In addition, we developed a new cuproptosisrelated molecular signature with great predictive performance for BCRFS and tumor immune environment using six DERRGs (including CALML5, MMP11, UBE2C, ANPEP, TMEM59L, COMP). These findings would be conducive to a deeper cognition of the potential mechanism of cuproptosis of PCa.

Gliomas are the most common and heterogeneous primary brain tumors, which leads to poor prognosis in many cases. Cellular senescence plays a key role in tumor progression and drug resistance, yet the prognostic value of senescence in gliomas remains unclear. Here, we identified key senescence-related genes through consensus clustering and weighted gene co-expression network analysis (WGCNA), and developed a cellular senescence-related gene prognostic signature (CSRGPS) using ten machine learning algorithms. The CSRGPS demonstrated strong predictive power, outperforming traditional clinical and molecular models. It stratified patients into distinct prognostic groups exhibiting differences in survival, clinical features, biological functions, and the tumor microenvironment. Single-cell analysis revealed a transition from low to high CSRGPS states. Furthermore, clinical data indicated an association between low CSRGPS and better outcomes following anti-PD-1 therapy. We also developed a nomogram integrating CSRGPS and clinical data, which further improved individualized prognosis prediction. Overall, CSRGPS offers a robust, clinically applicable tool for glioma prognosis and immunotherapy guidance, with potential utility in other cancers.

BackgroundPancreatic cancer (PC) exhibits an extremely poor prognosis due to its high heterogeneity. The senescence-associated secretory phenotype (SASP), a distinct secretory profile displayed by senescent cells, has been increasingly studied. However, the role of SASP in PC prognosis and treatment remains unclear.MethodsTranscriptomic sequencing data from PC patients were analyzed using consensus clustering based on SASP genes. A prognostic signature was subsequently constructed via Least Absolute Shrinkage and Selection Operator (LASSO) regression using survival-related SASP genes. Pathway enrichment analysis for distinct subgroups was performed using Gene Set Variation Analysis (GSVA). Comprehensive analyses of mutational landscapes and tumor immune microenvironments were conducted across risk-stratified PC samples.ResultsConsensus clustering based on SASP genes identified two SASP-associated clusters (SASPclusters), with cluster B demonstrating significantly worse prognosis than cluster A. Thirty-three SASP genes showed significant associations with PC prognosis, and a 7-gene SASP-based prognostic signature was established. High-risk patients exhibited significantly higher mutation rates. Distinct immune cell infiltration patterns, immune functions, checkpoint expression levels, and chemosensitivity profiles were observed between risk groups. Besides, we found that ANGPTL4 could promote PC cell proliferation, migration, and invasion.ConclusionMolecular subtyping and risk stratification based on SASP genes effectively predict PC prognosis and reveal heterogeneity in mutational burden, immune microenvironment, and therapeutic sensitivity. These computational findings deepen our understanding of potential role of SASP in PC and provide a theoretical foundation for personalized treatment strategies.

BackgroundImmunosuppressive breast cancer subtypes driven by regulatory T cells (Tregs) remain under-characterized, limiting precise identification of patients who may benefit from immunomodulatory therapies. Tregs are key mediators of immunosuppression within the tumor microenvironment (TME) and are closely associated with resistance to immune checkpoint inhibitors (ICIs). Therefore, defining and characterizing tumors with predominant Treg-mediated immunosuppression is essential for optimizing the use of Treg-targeted and combination immunotherapies.MethodsWe applied an unsupervised multi-omics integration approach across four molecular layers — mRNA, miRNA, DNA methylation, and proteomics —to identify immunologically distinct subtypes of breast cancer. Autoencoder-based dimensionality reduction followed by consensus clustering revealed a subgroup characterized by high Treg infiltration and immunosuppressive signaling, referred to as the Treg-enriched subtype. To evaluate therapeutic strategies, we employed a spatial quantitative systems pharmacology (spQSP) model simulating tumor–immune dynamics and tested Treg-targeted and PD-1 blockade therapies both alone and in combination. In vivo efficacy studies were conducted using the EMT6 syngeneic breast tumor model, characterized by an immunosuppressive tumor microenvironment, assessing the antitumor effects of a CCR8-targeted small molecule (IPG7236) as monotherapy or in combination with anti–PD-L1 treatment.ResultsThe C2 cluster exhibited elevated Treg-related signatures and a highly immunosuppressive tumor microenvironment. A similar Treg-enriched cluster was also identified in an independent cohort, supporting the robustness and clinical relevance of this immunosuppressive subtype. In-silico simulations performed under a C2-like, immunosuppressive context predicted that combining Treg-targeted therapy with PD-1 blockade would substantially enhance immune activation and tumor control compared with monotherapy. To experimentally validate these predictions, combination treatment of a CCR8 inhibitor (IPG7236) and anti–PD-L1 antibody demonstrated greater tumor growth inhibition than either monotherapy in the EMT6 model, confirming the predicted therapeutic synergy in Treg-enriched, immune-suppressive tumors.ConclusionThis study identifies Treg-enriched and immunosuppressive breast cancer subtype through integrative multi-omics analysis and demonstrates, through both in-silico and in-vivo approaches, the therapeutic potential of combining Treg-targeted and PD-L1 blockade therapies. These findings highlight Treg-mediated immunosuppression as a key determinant of therapeutic responsiveness, providing a biological rationale for patient stratification and guiding the development of personalized combination strategies for clinical translation.