Endometrial cancer (EC) is the most common gynecologic malignancy, with a steadily increasing incidence worldwide. Abnormal vaginal bleeding, a hallmark symptom, enables early diagnosis, which is critical for improving clinical outcomes. Pelvic magnetic resonance imaging (MRI) serves as the primary imaging modality for EC evaluation, offering detailed visualization of endometrial and myometrial invasion.This study proposes DCS-Net, a multi-task deep learning framework for the automated detection and staging of early-stage EC in MRI. The framework incorporates an advanced object detection module to accurately localize and crop the uterine region, followed by a convolutional neural network for staging classification.Experimental results show that the detection module achieves high localization performance, and the classification network reaches an accuracy of 90.8%. The region-focused approach improves staging accuracy by 5% compared to direct classification using unprocessed images.These results underscore the potential of DCS-Net to improve diagnostic efficiency and accuracy in alignment with clinical workflows. Future work will explore the integration of multi-parametric MRI data to further enhance diagnostic performance and address broader clinical needs.

Gastric cancer (GC) is a common malignancy of the digestive system, characterized by high invasiveness and metastasis, making it a leading cause of cancer-related deaths. Long non-coding RNAs (lncRNAs) play a crucial role in various types of cancer. This study aimed to elucidate the function of LINC01615 and its potential regulatory mechanisms in GC. In tissue and serum samples from GC patients, LINC01615 expression was significantly elevated and was associated with recurrence, distant metastasis, and TNM stage. LINC01615 promoted the proliferation, migration, and invasion of GC cells, and accelerated the growth of subcutaneous tumors in nude mice. Mechanistically, LINC01615 activated the WNT2/β-Catenin pathway by preventing WNT2 mRNA degradation and enhancing the nuclear translocation of β-Catenin. Furthermore, we identified the transcription factor YY1 as a key regulator of LINC01615 expression, which directly binds to its promoter region. Analysis of clinical samples revealed a positive correlation between the YY1/LINC01615/WNT2 signaling axis and poor prognosis in GC patients. Taken together, this study uncovers that the YY1/LINC01615/WNT2 signaling axis plays a crucial role in GC progression and may serve as a novel diagnostic marker and therapeutic target.

This study proposes a dual-branch framework for precise classification of breast tumor cellularity via histopathological images where it integrates two distinct branches: the Embedding Extraction Branch (embedding-driven) and the Vision Classification Branch (vision-based). The Embedding Extraction Branch uses the Virchow2 transformation to generate dense, structured embeddings, whereas the Vision Classification Branch employs Nomic AI Embedded Vision v1.5 to process image patches and produce classification logits. Both branches' outputs are combined to form the final classification. The framework also suggests Knowledge Block with fully connected layers, batch normalization, and dropout to improve feature extraction and reduce overfitting. The proposed approach reports high performance metrics, with an accuracy of [Formula: see text], specificity of [Formula: see text], and sensitivity, precision, and F1 score of [Formula: see text]. Also, ablation studies show the mandatory role of the embedding extraction branch; as its removal drastically reduces accuracy to [Formula: see text]. Furthermore, the Vision Classification Branch contributes significantly and its removal aims to a smaller decrease in the accuracy performance ([Formula: see text]). Additionally, data augmentation improves model performance and its exclusion results in a notable decline in accuracy performance ([Formula: see text]). The approach's robustness is validated through statistical analysis that reports low variance and high consistency across multiple performance metrics.

The study aimed to explore the distribution of circulating tumor cells (CTCs), circulating tumor endothelial cells (CTECs), and their subtypes in non-cancer and bladder cancer individuals, focusing on their prognostic value in high-risk non-muscle invasive bladder cancer (NMIBC). Researchers analyzed 59 fresh peripheral blood samples using subtraction enrichment and immunostaining fluorescence in situ hybridization (SE-iFISH). Samples were collected from healthy individuals (n = 18), patients with benign urinary conditions (n = 2), newly diagnosed bladder cancer patients (n = 20), and NMIBC patients after repeated transurethral resection of bladder tumor (R-TURBT) (n = 19). NMIBC patients had significantly higher total CTCs. In newly diagnosed bladder cancer patients, large CTCs constituted 58.8%. The most common karyotype was ≥ pentaploid CTCs (61.2%). In the non-cancer group, large CTCs constituted 83.0%, with ≥ pentaploid CTCs comprising 72.3% of aneuploid CTCs. For NMIBC patients after R-TURBT, those without recurrence had 16% small CTCs. Conversely, the recurrence group had 71% small CTCs, where tetraploid CTCs were predominant (40%). By performing logistic ridge repression, the ≥ pentaploid small CTC is noted as an important indicator of recurrence. The presence and proportion of small CTCs can serve as a prognostic marker in NMIBC patients following R-TURBT, potentially guiding patient management and surveillance strategies.

The bone marrow (BM) is a complex and compartmentalized tissue where spatial context plays a critical role in regulating cell behavior, signaling, and disease progression. To capture these dynamics, we apply spatial transcriptomics using the Visium Spatial Gene Expression platform on formalin-fixed paraffin-embedded (FFPE) BM sections from both healthy and Multiple Myeloma (MM) mouse models, as well as MM patient samples. Overcoming the technical challenges of working with mineralized long bone tissue, we develop a custom analytical framework integrating spatial and single-cell transcriptomic data to map cellular composition and interactions in situ. This approach enables the spatial characterization of transcriptionally heterogeneous malignant plasma cells (MM-PC) and their surrounding microenvironments. We identify spatially distinct gene programs linked to MM pathogenesis, including signatures of NETosis and IL-17 signalling, which are reduced in MM-PC-rich regions. Additionally, a transition gradient from effector to exhausted T cell phenotype is associated with increased remoteness from MM-PC. These spatial patterns are identified in FFPE BM biopsies from MM patients with varying tumor burdens. In summary, our study demonstrates both the capabilities and limitations of Visium technology in characterizing spatially regulated mechanisms underlying MM pathogenesis.

Apoptosis is a form of programmed cell death mediated by caspases and regulated through both intrinsic and extrinsic pathways. Many cancers evade apoptosis, resulting in uncontrolled cell survival and tumor progression. Inhibitor of apoptosis proteins (IAPs), including XIAP, cIAP1, cIAP2, ML-IAP (also known as Livin), Survivin, NAIP, ILP-2, and BRUCE (also known as Apollon), play crucial roles in suppressing apoptosis and promoting cancer cell survival. Survivin, which is overexpressed in various malignancies, inhibits caspase activity, protects XIAP from proteasomal degradation, and suppresses the intrinsic apoptotic pathway by inhibiting caspase-9 activity. In this study, we investigated the protective role of Survivin on other IAPs, particularly NAIP and cIAP1, using bioinformatics approaches such as homology modeling, molecular docking, and molecular dynamics simulations. Experimental validation in MCF-7 breast cancer cells demonstrated that a novel anticancer peptide, P3, disrupts the interaction between Survivin and IAPs. At 25 µM, P3 significantly enhanced caspase-8 and -9 (initiator) and caspases-3 and -7 (executioner) activities, activating pathways. These effects were confirmed by flow cytometry and DAPI/PI staining, showing increased apoptosis without necrosis. Our findings indicate P3 as a promising peptide-based therapy to overcome apoptosis resistance in breast cancer by targeting Survivin-IAP complexes, providing a foundation for novel anticancer strategies.

The gastrin-releasing peptide receptor (GRPR) is expressed across multiple human cancer types, including head and neck tumors, and has also been detected in the oral mucosa of patients bearing head and neck neoplasms. This study aimed to evaluate GRPR expression in tumor samples obtained from patients with histopathologically confirmed squamous cell carcinomas of the head and neck and esophagus. GRPR protein expression was analyzed by immunohistochemistry in 80 tumor specimens and 10 cancer-free tissue samples. GRPR positivity was observed in 72.5% of tumors, whereas all healthy control samples were negative. Additionally, an expression threshold of 10% was sufficient to distinguish between positive and negative tumors (P < 0.0001). Overall survival for the cohort was 64.7%. GRPR expression tends to influence patient outcomes, with an expression level exceeding 50% correlating with worse survival (p = 0.031). GRPR is overexpressed in head and neck and esophageal squamous cell carcinoma, and higher expression was associated with worse survival, suggesting a promising role of GRPR as a new biomarker for early diagnosis and prognosis.

Pulmonary lymphangioleiomyomatosis (LAM) is a metastatic sarcoma, but the mechanisms of LAM metastasis are unknown. Extracellular vesicles (EVs) regulate cancer metastasis; however, their roles in LAM have not yet been thoroughly investigated. Here, we report the discovery of distinct LAM-EV subtypes derived from primary tumor or metastasizing LAM cells that promote LAM metastasis through ITGα6/β1-c-Src-FAK signaling, triggered by the shuttling of ATP synthesis to cell pseudopodia or the activation of integrin adhesion complex, respectively. This signaling leads to increased LAM cell migration, invasiveness, and stemness and regulates metastable (hybrid) phenotypes that are all pivotal for metastasis. Mouse models corroborate in vitro data by demonstrating a significant increase in lung metastatic burden upon exposure to EVs through distinct mechanisms involving either lung resident fibroblasts or metalloproteinases' activation that are EV subtype dependent. The clinical relevance of these findings is underscored by increased EV biogenesis in LAM patients and the enrichment of these EV cargo with lung-tropic integrins and metalloproteinases. These findings establish EVs as a novel therapeutic target in LAM, warranting future clinical studies.

Plasticity, or the ability to rapidly and reversibly change phenotypes, may help explain how a single progenitor cell eventually generates a tumor with many different cell phenotypes. Normal colon plasticity is characterized by a conserved and broadly permissive epigenome, where expression and phenotype are determined by the microenvironment instead of epigenetic remodeling. To determine whether this stem-like plasticity is retained during progression, gene expression was measured with spatial transcriptomics and compared with gene-level DNA methylation in two colorectal cancers (CRCs). Like normal colon, genes that were differentially expressed between regions, subclones, and phenotypes (superficial, invasive, and metastatic) tended to have lower DNA methylation variability. We propose a quantitative signal of plasticity that correlates gene epigenetic variability with gene expression variability. In this framework, negative correlation implies phenotypic plasticity, as more variably expressed genes tend to have less epigenetic variability. We verify the presence of this signal in multiple external single-cell RNA-Seq datasets, in both normal colon and CRC samples. Therefore, the plasticity of normal colon appears to be retained during progression. A CRC progenitor with a preconfigured plastic phenotype is poised for rapid growth because it expresses, as needed, transcripts required for progression with minimal epigenetic remodeling.

This study was designed to illustrate the mechanism underlying the up-regulation of LINC00857 expression and to identify novel molecular targets for the treatment of ovarian cancer (OC). The LINC00857 and methyltransferase-like 3 (METTL3) expression levels were observed in clinical OC tissues and cells, and their correlation within tissues was analyzed. To further explore the relationship between LINC00857 and METTL3, a combined transfection of pcDNA3.1-LINC00857 and METTL3 siRNA was performed. Subsequently, cell invasion, viability, migration, and sphere-forming capabilities were assessed using Transwell, Cell Counting Kit-8, scratch and sphere-formation assays. Furthermore, western blot analysis was conducted to determine the expression of proteins related to cancer cell stemness and the yes-associated protein (YAP) pathway. The relationship between LINC00857 and METTL3 was verified using the MeRIP-qPCR kit, RNA pull-down assay and RNA stability assay. Both LINC00857 and METTL3 demonstrated high expression levels in OC cells and tissues, with a positive correlation observed in clinical tissues. When knocking down the LINC00857 expression level, the invasion, migration, proliferation, and sphere-forming capabilities of cells were all notably reduced. Knocking down LINC00857 expression also markedly decreased the activity of the YAP pathway and the expression of proteins related to cancer cell stemness. Overexpression of LINC00857 yielded opposite effects. When knocking down METTL3, the stability of LINC00857 and the modification level of N6-methyladenosine were remarkably decreased. Moreover, there was interaction between METTL3 and LINC00857. METTL3-mediated N6-methyladenosine modification of LINC00857 enhanced metastasis and stemness of OC cells via activating the YAP-TEA domain transcription factor pathway.

Oxidative phosphorylation (OXPHOS) is a key player in mitochondrial bioenergetic functions. In hepatocellular carcinoma (HCC), OXPHOS slows down or switches to glycolysis via what is known as the Warburg effect. The altered respiration in cancer was reported to affect mitochondrial temperature. We investigated the impact of the metabolic switch on the mitochondrial temperature in HepG2 HCC cell line. Metformin (N, N-dimethylbiguanide) treatment was used to suppress glycolysis to emulate lower metabolically active cells (Met-HepG2). The mitochondrial temperature was assessed using mito-thermo yellow (MTY) absorbing mitochondrial radiant heat. Mito-tracker green (MTG) fluorescent dye was used to confirm mitochondrial localization. Our data showed lower MTY dye intensity in the Met-HepG2 treated group, indicating a significant increase in mitochondrial temperature compared to untreated HepG2 cells (NT-HepG2). Genotypic analysis of the metabolic respiration gene expression showed significant down-regulation in glycolytic genes (ERR-gamma, HK2, PGK, ALDOC, TPI1, IDH1, and PKM2) in the Met-HepG2 cells compared to the NT-HepG2 cells. OXPHOS as evidenced by ATP, ROS, and NADPH production was significantly up-regulated in the Met-HepG2 group compared to the NT-HepG2 group. Transmission electron microscopy showed fewer mitochondria with swollen elongated appearance, as a marker for activated OXPHOS in the Met-HepG2 group. These data show a correlation between HepG2 altered metabolism and mitochondrial temperature and suggest that less metabolically active HepG2 cells are correlated with higher mitochondrial temperature, providing evidence for a possible role of mitochondrial temperature in diagnosis of HCC.

The Tumor-Stroma Ratio (TSR) is a critical prognostic factor in colorectal cancer (CRC), offering insights into tumor microenvironment interactions. However, traditional TSR assessment methods are subjective and labor-intensive. This study is among the first approaches that introduce a novel integrative deep learning to leverage Transformer mechanisms for enhanced spatial context understanding in TSR assessment for CRC, addressing limitations in conventional CNN-only models and combining. Our methodology involves classifying patch images into normal and abnormal categories using a CNN-based model, then segmenting abnormal patches into tumor and stroma areas in whole-slide images (WSIs) through a hybrid CNN-Transformer UNET model. TSR is quantitatively assessed based on pixel area, providing a robust and objective measure of the tumor microenvironment. The classification model performed superiorly by achieving an overall classification accuracy of 93.53%, precision of 0.9040%, recall of 0.9644%, f1-score of 0.9332%, and Matthew's correlation coefficient (MCC) of 0.8724, sufficient to maintain consistent TSR evaluation in case of eventual misclassification. In segmentation tasks, the proposed Efficient-TransUNet model achieved the highest Aggregated Dice Coefficient (ADC) of 0.938 for stroma and 0.921 for tumor class, outperforming existing models. These results underscore the potential of this hybrid deep learning framework to enhance the precision of pathological evaluations and contribute to improved clinical outcomes.

Colorectal cancer (CRC), a highly heterogeneous disease, is the second leading cause of cancer related deaths. Mounting evidence has indicated that the zinc finger of the cerebellum 5 (ZIC5) is involved in the pathological processes of cancer. At present, there are few and conflicting studies on ZIC5 in CRC. The aim of this study was to elucidate the clinical value, function and molecular mechanism of ZIC5 in colon cancer. ZIC5 expression analysis in pan-cancer was performed using the TCGA database. The correlation of ZIC5 expression with clinicopathologic parameters of colon cancer was assessed in public and external datasets, respectively. The influence of ZIC5 expression on survival was analyzed by Kaplan-Meier (KM) curve and Cox regression models. In addition, a series of in vitro experiments on cell proliferation, invasion, migration and stemness were performed to further explore the function and potential mechanisms of ZIC5 in colon cancer. ZIC5 mRNA expression is significantly elevated in the vast majority of cancers. We found that ZIC5 is elevated in colon cancer tissues, and high ZIC5 expression was associated with poorer survival of colon cancer patients. ZIC5 expression was positively correlated with the malignant pathological phenotype, and it's an independent risk marker for the overall survival (OS) of colon cancer patients. We demonstrated that ZIC5 promoted proliferation, invasion, migration and EMT of colon cancer cells. Additionally, ZIC5 enhances colon cancer stem cell (CSC) properties. Mechanistically, ZIC5 functions in colon cancer cells by upregulating Wnt signaling. ZIC5 could enhance Wnt/β-catenin signaling by interacting with β-catenin. Interestingly, in HCT116 cells, ZIC5 doesn't affect the protein level of β-catenin and its nuclear translocation likely due to the deletion mutation of β-catenin at Ser45. We identified ZIC5 as a survival marker for colon cancer patients. Our study shows that inhibition of ZIC5 might be a potential therapeutic strategy for colon cancer.

Breast cancer (BC) is the most common malignancy among women, with its progression and prognosis significantly influenced by the tumor microenvironment (TME). Age-related differences in TME composition lead to distinct tumor behaviors: young patients (≤ 40 years) exhibit aggressive tumors, while elderly patients (> 70 years) experience immunosenescence and reduced therapy responses. We performed single-cell RNA sequencing (scRNA-seq) analysis on tumors from 10 breast cancer patients (5 ≤ 40 years, 5 ≥ 70 years), encompassing 33,664 high-quality cells. After cell annotation and batch correction, malignant epithelial cells were identified using inferCNV. We applied pseudotime trajectory analysis, pathway enrichment, and cell-cell communication profiling to investigate age-specific TME dynamics. Survival relevance was assessed using a GEO cohort (GSE20685) of young breast cancer patients, and immunohistochemical staining was performed on clinical tumor and fibroadenoma tissues to validate protein-level expression of key ISGs. In young patients, malignant epithelial cells showed gradual upregulation of interferon-stimulated genes (ISGs) such as IFI44, IFI44L, IFIT1, and IFIT3 along the pseudotime trajectory, suggesting their involvement in early tumorigenesis. High expression of these ISGs was significantly associated with poor overall survival in a young BC cohort (GSE20685). Immunohistochemical validation further confirmed elevated IFIT3 protein levels in young tumor tissues. In contrast, elderly patients had a TME enriched in macrophages and fibroblasts, with activation of immunosuppressive pathways (e.g., SPP1, COMPLEMENT). Our integrative analysis identifies ISGs as key transcriptional drivers of tumorigenesis in young breast cancer, with potential prognostic and therapeutic value. Despite limited sample size, the combination of single-cell transcriptomics, clinical survival data, and protein-level validation provides robust evidence of age-specific TME remodeling. These findings support the development of age-tailored immunotherapy strategies targeting interferon signaling in young patients and immune checkpoint pathways (e.g., LAG3, CTLA4) in elderly individuals.

This study introduces a unique magnetic resonance imaging dataset focusing on metastatic breast cancer to the brain, a significant clinical challenge in cancer treatment. Comprising 297 T1-weighted post-contrast images from 165 patients, this dataset from the University of Minnesota Medical Center is the first dedicated to breast cancer brain metastases. This collection includes expert-reviewed lesion segmentations with original image files, genetic markers, and an extensive array of tumor-derived radiomic features. The dataset's uniqueness lies in its detailed focus on metastatic breast cancer to the brain-offering a rich resource for advanced image-based tumor phenotyping and the vast potential for radiogenomic-based predictions based on machine learning model development. The inclusion of clinician-reviewed tumor segmentations and radiomic features, encompassing shape and texture characteristics, enhances the dataset's utility. This dataset aims to facilitate a deeper understanding of breast cancer metastasis to the brain, promote advancements in precision medicine, and improve patient care.

Head and neck squamous cell carcinoma (HNSCC) exhibits a poor prognosis, with 5-year survival rates below 50%. This study employed network pharmacology to investigate the anti-HNSCC mechanism of silibinin, a plant-derived compound with established anticancer activity. We obtained potential silibinin targets from pharmacological databases and HNSCC-associated targets from TCGA. We performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses to identify critical pathways and biological processes. Through protein-protein interaction (PPI) network screening, we selected hub genes for molecular docking validation. We evaluated silibinin's effects on HNSCC proliferation and invasion using CCK-8 assays, colony formation tests, and cell invasion experiments. Our data suggested that silibinin may inhibit HNSCC progression through modulation of the interleukin-17 signaling pathway. Molecular docking confirmed strong binding affinity between silibinin and key targets, supporting its potential as an HNSCC therapeutic agent.

STX16 has been studied in various cancers, where it is suggested to regulate tumor cell proliferation, migration, and invasion by affecting vesicle trafficking and signal transduction pathways. However, its specific role in clear cell renal cell carcinoma (ccRCC) remains unclear. This study aims to systematically investigate the expression patterns, clinical significance, prognostic value and functional role of STX16 in ccRCC. Bioinformatics analyses using TCGA, CTPAC, TIMER, and UALCAN databases evaluated STX16 expression, clinical correlations, prognosis, and immune infiltration. Functional enrichment, co-expression, and PPI analyses explored STX16-associated pathways. Single-cell sequencing elucidated tumor microenvironment heterogeneity. Laboratory validation included Western blot, immunohistochemistry, and functional assays with siRNA-mediated STX16 knockdown in ccRCC cell lines. STX16 was significantly upregulated at mRNA and protein levels in ccRCC tissues. High STX16 expression correlated with advanced tumor stages, poor overall survival (OS), and disease-specific survival (DSS). Multivariable Cox regression identified STX16 as an independent prognostic factor. STX16 influenced immune infiltration, particularly involving CD4 + T cells, macrophages, and neutrophils, and was associated with immune pathways. Single-cell sequencing revealed heterogeneous STX16 expression across tumor microenvironment cell types. STX16 knockdown inhibited proliferation, migration, and invasion of ccRCC cell lines. STX16 is upregulated in ccRCC and acts as an independent prognostic factor associated with poor outcomes. Its role in promoting tumor progression and modulating immune infiltration highlights STX16 as a potential biomarker and therapeutic target in ccRCC.

Although it is the most common subtype of lung cancer in clinical practice, lung adenocarcinoma (LUAD) was proven to be associated with a poor prognosis. In recent years, lactate metabolism has been considered an important biological mechanism in lung cancer. However, the mechanisms of lactate-related genes in tumour microenvironment (TME) of LUAD are unknown. Lactate-related genes modification patterns in 701 LUAD samples from TCGA and GEO database were systematically assessed on the basis of 16 lactate-related genes (LRGs).We also identified the correlation of these clusters to the TME and immune cell infiltration in the TME. After unsupervised clustering analysis was performed, the LUAD samples were divided into three lactate-related gene phenotypes on the basis of 36 prognostic lactate-related genes (PLRGs). These molecular subtypes have different immune cell infiltration characteristics and pathway enrichment. Using LASSO, a 16-LRG risk signature was constructed. The lactate-related signature demonstrated a stable and accurate ability to predict the prognosis of LUAD in patients. Furthermore, the lactate-related signature demonstrated good predictive ability for immune infiltrating cells, tumour mutation burden, and response to immunotherapy. The 16-LRG risk signature was subsequently verified in the GSE50081 GEO cohort. This study revealed the significant clinical utility of the 16-LRG risk signature in the understanding the TME and prognosis of LUAD. The 16-LRG risk signature is conducive to understanding immune cell infiltration in the LUAD TME and contributes to the selection of more effective immunotherapy strategies.

Lung cancer remains the leading cause of cancer-related mortality worldwide, largely due to late-stage diagnoses that severely limit therapeutic interventions. In this context, nanoparticle-mediated photothermal therapy (PTT) has emerged as a promising and minimally toxic modality for solid tumors. We synthesized gold nanoparticles (AuNPs) with three distinct morphologies-spheres, rods, and stars-and functionalized them with polyethylene glycol (AuNPs-PEG) or polyethylene glycol conjugated with 2-deoxy-D-glucose (AuNPs-Gluc). In vitro analyses using human (A549, H1299) and murine (LLC) lung carcinoma cell lines demonstrated that PEGylation significantly attenuated AuNP-associated cytotoxicity, while glucose functionalization further enhanced biocompatibility. Inductively coupled plasma mass spectrometry quantification confirmed superior cellular uptake of AuNPs-Gluc compared to AuNPs-PEG (p < 0.05). Subsequent irradiation with a 980 nm diode laser (1 W) induced robust thermal damage and apoptotic cell death selectively in cancer cells treated with AuNPs-Gluc, sparing non-tumoral cells. Among the morphologies tested, star-shaped AuNPs exhibited the highest photothermal efficiency. In vivo experiments further substantiated the therapeutic potential, as combined administration of AuNPs-Gluc and laser irradiation significantly suppressed tumor growth (p < 0.01). Collectively, these findings highlight the utility of glucose-functionalized AuNPs as effective vectors for targeted PTT in lung cancer, supporting their translational relevance for future clinical applications in advanced-stage disease.

Mitochondrial autophagy (mitophagy) in macrophages is crucial yet poorly understood within the lung adenocarcinoma (LUAD) tumor microenvironment. This study aimed to identify key macrophage mitophagy-related genes and develop a robust prognostic model for LUAD patients. By integrating single-cell transcriptomics with machine learning algorithms, including LASSO, SVM, and random forest, we identified TUBB6 and CAT as core prognostic genes. A novel risk model based on these genes (RiskScore = 0.225 × TUBB6 - 0.19 × CAT) was constructed and validated, demonstrating that patients in the high-risk group had significantly shorter overall survival (P < 0.001). The high-risk score also correlated with an altered immune microenvironment and increased sensitivity to chemotherapies like cisplatin and gemcitabine. Furthermore, in vitro experiments confirmed that macrophage-specific overexpression of TUBB6 significantly enhanced LUAD cell proliferation, migration, and invasion through secreted factors. Our findings establish a reliable, macrophage mitophagy-based prognostic model and highlight TUBB6 and CAT as novel biomarkers and potential therapeutic targets, offering new avenues for precision medicine in LUAD.