This study investigated the potential of enhancer RNAs (eRNAs) as biomarkers for breast cancer, complementing traditional protein-coding transcripts. By profiling eRNA expression across over 1000 TCGA-BRCA samples and healthy tissues, researchers developed a 10-eRNA signature that effectively stratified patients into high- and low-risk groups, demonstrating independent prognostic value. Furthermore, a 19-eRNA panel was designed for early detection, showing excellent performance on an independent cohort. These tools have been integrated into eRNACare, a publicly accessible web application, highlighting the promising role of eRNAs for breast cancer risk stratification, therapeutic guidance, and early diagnosis.
This study investigated mutations in the untranslated regions (UTRs) of EGFR, AKT1, and HRAS genes in Pakistani colorectal cancer patients. Whole-exome sequencing and extensive in silico analysis identified several UTR mutations. These alterations demonstrated significant functional impacts, including the loss or gain of transcription factor binding sites, changes in mRNA stability, and the loss of oncomiR binding sites. The findings suggest that these UTR mutations could dysregulate the expression of these oncogenes and contribute to colorectal cancer progression.
The ITALUNG trial, a randomized controlled study, evaluated low-dose computed tomography (LDCT) for lung cancer screening in over 3100 high-risk individuals. This study also analyzed the performance of a biomarker panel (loss of heterozygosity/microsatellite instability and circulating free DNA) in blood and sputum samples. The panel demonstrated high sensitivity for detecting lung cancers, even in pre-diagnostic samples collected several years before clinical diagnosis. These biomarkers may be associated with carcinogenic field effects and early tumor presence.
This retrospective study investigated Claudin 18.2 (CLDN18.2) expression via immunohistochemistry in 64 patients with advanced gastrointestinal peritoneal disease. The aim was to assess CLDN18.2's potential as a therapeutic target beyond upper gastrointestinal cancers. Results revealed significant CLDN18.2 expression in a subset of appendiceal peritoneal metastases, particularly in high-grade appendiceal mucinous carcinomas (22% strong positivity). In contrast, expression was low in colorectal adenocarcinomas. These findings suggest CLDN18.2 could be a promising therapeutic target for patients with appendiceal mucinous carcinomas, potentially offering new options where curative treatments are limited.
This study employed a deep learning-based graph neural network model to identify genes and pathogenic genetic variations driving gastric cancer pathogenesis. Researchers focused on tumor suppressor proteins exhibiting both dysregulated expression and high mutation propensity. The model identified 1,886 relevant genes, with TP53, ERBB2, CDH1, and APC emerging as prominent hub genes. Missense and nonsense variants in TP53, CDH1, and APC were validated, affecting post-translational modification sites and correlating with altered patient survival. The study also highlighted EZH2 and miR-129-5p as key regulators and the dysregulation of oncogenes like CCNE1/2 and FGFR2.
This study identified and validated novel promoter DNA hypermethylation signatures as prognostic biomarkers in oral squamous cell carcinoma (OSCC). Combining a discovery approach based on pharmacologic demethylation and RNA sequencing with multi-platform validation, five candidate genes (GPX3, ANG, CTGF, GPRC5B, BAMBI) were identified. Their promoter hypermethylation is associated with cancer-specific transcriptional and protein silencing. Notably, CTGF hypermethylation was independently associated with poor overall survival, and a combined ANG+CTGF signature demonstrated superior and reproducible prognostic performance. These findings pave the way for clinical risk stratification in OSCC patients.
The phase II DETECT IVb trial assessed the efficacy and safety of eribulin in patients with HER2-negative metastatic breast cancer, focusing on circulating tumor cell (CTC) dynamics. The study confirmed eribulin as an effective treatment option for this high-risk population, showing a median progression-free survival of 4.6 months and a median overall survival of 13.4 months. Furthermore, CTC clearance was significantly associated with improved progression-free survival. The observed adverse events were consistent with the known safety profile of eribulin.
This study investigated the stability of MGMT promoter methylation in recurrent glioma patients to inform treatment decisions. Researchers analyzed paired tumor samples from 426 patients, revealing that MGMT methylation status remained unchanged in 84% of cases. The 16% of observed changes primarily occurred in tumors with initial methylation levels near classification thresholds. These findings suggest that reassessing MGMT status at recurrence, with quantitative reporting, could improve therapeutic decision-making, particularly for temozolomide rechallenge.
This retrospective study investigated the spatial heterogeneity of PD-L1 and c-MET expression in oral squamous cell carcinoma (OSCC) by comparing primary tumors and matched lymph node metastases. Findings revealed that both PD-L1 and c-MET were frequently expressed in primary tumors but at significantly lower levels in corresponding lymph node metastases. A substantial discordance in PD-L1 CPS status was observed between the two tumor compartments, potentially impacting biomarker classification. While higher PD-L1 expression in primary tumors was associated with nodal metastasis, no significant correlation was found between PD-L1 and c-MET. These results highlight the critical role of spatial heterogeneity in biomarker assessment for OSCC.
A machine learning-based clinicogenomic model has been developed and validated to improve prognostic stratification in ER-positive/HER2-negative early breast cancer. This model integrates a research-grade implementation of the 21-gene recurrence score, a 14-gene immunoglobulin signature, and clinicopathologic features. Validated across four independent datasets totaling 5132 patients, it demonstrated improved discrimination and identified larger low-risk groups compared to the standalone 21-gene recurrence score. Explainability analyses revealed time-dependent feature contributions, highlighting the potential of this integrative approach for more accurate risk assessment.
This prospective study evaluated the prognostic significance of KRAS-mutant circulating tumor DNA (ctDNA) in patients with localized pancreatic ductal adenocarcinoma (PDAC). Analyses compared detection by next-generation sequencing (NGS) and digital droplet PCR (ddPCR) at various stages of perioperative treatment. Results indicated that the more sensitive ddPCR detected KRAS-mutant ctDNA in a greater proportion of patients than NGS. Detection by either method was associated with shorter overall survival, with ddPCR providing superior prognostic discrimination by identifying additional patients with poor outcomes missed by NGS. Integrating both methods could enhance perioperative risk stratification.
This study analyzed a cohort of 1,254 patients with biliary tract cancer to assess the utility of targeted NGS molecular profiling. It revealed that 59% of patients harbored at least one OncoKB alteration, and 32.2% had level 1/2 alterations, with varying frequencies across tumor locations. Targeted therapies were associated with improved progression-free survival but not overall survival. The study also identified co-occurring alterations and acquired resistance mechanisms, such as ERBB2 loss or alterations in the RAS, MEK, MET, MYC, and CDKN2A pathways. These findings highlight genomic heterogeneity and the importance of precision medicine in informing future drug development efforts.
This study conducted an integrated multi-omic analysis of 214 pediatric adrenocortical tumors, identifying four distinct risk groups. A high-risk subgroup is characterized by CpG island hypermethylation, chromosomal instability, and dismal survival. These tumors exhibit transcriptional co-activation of WNT signaling and AP-1 transcriptional programs, alongside a balanced admixture of zona glomerulosa and fasciculata/reticularis-like cells. Entinostat, a histone deacetylase inhibitor, was shown to reverse promoter methylation, silence AP-1 activity, and induce apoptotic reprogramming in tumor models. These findings establish a molecular framework for risk stratification and identify actionable therapeutic vulnerabilities.
This study investigates the potential of seminal fluid as a non-invasive liquid biopsy for detecting clinically significant prostate cancer. Seminal fluid samples were collected at home from 279 men scheduled for prostate biopsy, and their cell-free DNA (cfDNA) underwent methylation analysis. Researchers characterized seminal cfDNA, identifying granulocytes and sperm as primary contributors, but confirming the detectable presence of prostate-derived cfDNA. Fragment-level methylation features were analyzed to identify signals associated with clinically significant prostate cancer, highlighting the promise of this approach for diagnosis and monitoring.
Pirmoradi S, Vaghefi Moghaddam S, Ardalan M, et al.
PloS one ·
This study introduces an artificial intelligence framework utilizing graph neural networks (GNNs) for prostate cancer risk stratification, integrating mRNA expression profiles with functional interaction networks. The aim is to classify patients into low-, medium-, and high-risk groups, defined by Gleason scores, to guide treatment decisions and prevent overtreatment. Developed using TCGA data, the model achieved an overall accuracy of 80% and high AUCs on an independent test set. Differentially expressed genes were identified for each risk group, highlighting the molecular heterogeneity of disease progression.
This article describes the development and validation of a plasma cell-free DNA (cfDNA) fragmentomics-based machine learning classifier for differentiating benign and malignant renal masses. The model, trained on 331 participants and validated on 144, integrates fragmentation features like copy number variation (CNV), fragmentation-based methylation (FRAGMA), and nucleosome footprint (NF) derived from low-pass whole-genome sequencing. It demonstrated strong discriminative performance with AUCs of 0.956 and 0.946 in the training and validation cohorts, respectively. This noninvasive classifier maintains robust performance even with low tumor fractions and could improve clinical decision-making for indeterminate renal lesions.
This retrospective study, leveraging the phase III SUCCESS-A trial, assessed the prognostic value of tissue-free circulating tumor DNA (ctDNA) detection in patients with early breast cancer. Plasma samples were analyzed approximately two years after adjuvant chemotherapy in 313 patients. ctDNA detection, observed in 5.8% of patients, was strongly associated with a significantly increased risk of distant recurrence and shorter overall survival. ctDNA was detected a median of 7.9 months before recurrence, showing a sensitivity of 73% and a specificity of 99.6% for distant recurrence. These findings suggest that post-treatment ctDNA can stratify patients at high risk of recurrence.
This phase 1-2 study investigated the safety and efficacy of daraxonrasib, a multiselective RAS(ON) inhibitor, in patients with previously treated advanced RAS-mutant non-small-cell lung cancer (NSCLC). Patients received daily oral doses ranging from 10 to 400 mg. Results indicated that 99% of patients experienced adverse events of any grade, and 54% had grade 3 or higher events, with common side effects including rash and diarrhea. Antitumor activity, assessed by objective response rate, was over 30% across the evaluated dose groups.
This study investigated non-parametric differential methylation analysis to characterize histotype-specific promoter regions in early-stage epithelial ovarian cancer (EOC). The aim was to identify candidate epigenetic biomarkers for diagnosis and histotype stratification. Researchers applied this approach to an EOC cohort and compared results with existing programs. The identified promoter regions were highly reproducible and effectively stratified EOC histotypes in external cohorts through predictive classification. The study also revealed differences in methylation distribution types among histotypes, suggesting that methods sensitive to non-normally distributed data might be poorly suited for comparing groups with different distribution types.
This retrospective single-center study investigated the role of TERT promoter mutations as a biomarker to guide treatment in advanced melanoma, particularly in BRAF V600 mutant patients. TERT mutations were detected in 73% of 159 patients and frequently co-occurred with BRAF V600 mutations. While TERT status was not an independent prognostic factor, exploratory analyses suggested a potential interaction with treatment type. Patients with BRAF/TERT co-mutations showed longer overall survival when treated with immune checkpoint inhibitors, whereas the opposite trend was observed for TERT wild-type patients receiving BRAF/MEK inhibitors. These preliminary findings warrant further validation in prospective studies.
This study introduces GiGCN, a novel graph convolutional network framework designed for the joint identification of synthetic lethal (SL) and synthetic viable (SV) genetic interactions. By integrating verified genetic interactions and gene semantic similarities, GiGCN separately models distinct functional dimensions within gene networks. This approach significantly enhances prediction accuracy compared to existing methods, achieving AUC and AUPRC values of 0.978 and 0.944, respectively. Further analyses revealed SL interactions centered on the MYC oncogene and SV interactions linked to autophagy, providing crucial mechanistic insights. The ultimate goal is to facilitate the discovery of actionable therapeutic targets for cancer and other human diseases.
This study evaluated the utility of RNA sequencing and immunohistochemistry (IHC) for detecting NTRK fusions in thyroid carcinoma, particularly those potentially missed by gene panel testing. Two patient groups were analyzed: advanced thyroid cancer cases undergoing panel testing and surgical papillary thyroid carcinoma cases screened by IHC. Results indicated that a gene panel test missed one SQSTM1-NTRK3 fusion, while pan-TRK IHC showed variable sensitivity depending on the fusion type. The study concludes that active screening for NTRK fusions is desirable, especially in patients without MAPK pathway abnormalities or with positive pan-TRK staining.
This study introduces DNAmBERT, a Transformer-based deep learning model for non-invasive cancer diagnosis. It jointly models DNA sequence context and read-level methylation haplotype structure from cfDNA, addressing limitations of current computational methods. Evaluated across various platforms and cancer types, DNAmBERT achieved high performance with AUCs up to 0.99-1.00. The model supports multi-cancer and stage-aware classification, including early-stage disease, and demonstrates effective cross-cancer transfer learning.
This study characterized the molecular subtypes and the spectrum of POLE gene mutations in a South Indian cohort of endometrial carcinoma patients. Using the ProMisE classification, researchers identified four molecular subgroups: POLE ultra-mutated, MMR-deficient, p53-abnormal, and no specific molecular profile (NSMP). POLE gene sequencing revealed known variants like V411L, as well as several non-hotspot variants, six of which were unique to this cohort and four predicted to be oncogenic. p53-abnormal tumors exhibited a more aggressive phenotype and increased chemotherapy use. The integration of molecular profiling with histopathology is deemed crucial for enhancing endometrial carcinoma classification and prognostic assessment.
A recent study from the ORCHARD program provided the largest histogenomic characterization of first-line osimertinib resistance mechanisms. Utilizing dual-modality sequencing on both tissue and plasma samples, resistance alterations were identified in 87% of patients. Furthermore, polyclonal mechanisms were detected in 46% of patients, highlighting the intricate biological landscape post-osimertinib treatment. These findings prompt a re-evaluation of post-EGFR tyrosine kinase inhibitor therapeutic strategies, weighing targeted versus broadly acting regimens.
This SUMMIT basket trial study evaluated the efficacy of neratinib alone or in combination with trastuzumab in patients with HER2-mutant metastatic triple-negative breast cancer (TNBC). Results showed confirmed objective response rates of 40% for neratinib alone and 35.3% for the combination, with median progression-free survival times of 2.89 and 6.24 months, respectively. Circulating tumor DNA (ctDNA) monitoring correlated decreases in HER2 mutation variant allele frequencies with treatment response and identified the emergence of resistance mutations. The neratinib and trastuzumab combination appears to prolong responses compared to neratinib alone, offering a novel approach for biomarker-defined metastatic TNBC.
The Phase II ORCHARD study investigated resistance mechanisms to first-line osimertinib in patients with EGFR-mutated non-small cell lung cancer (NSCLC). Utilizing next-generation sequencing (NGS) on both tissue and plasma samples, the study characterized the molecular profile of tumors after progression. Resistance alterations were detected in 87% of samples, often multiple, involving genes such as TP53, MDM2/4, PI3K, SOX2, and MYC. These findings highlight the heterogeneity of resistance and provide a rationale for exploring combination therapies to improve patient outcomes.
This study conducted comprehensive genomic profiling on 39 cases of pleomorphic liposarcoma (PLPS) to characterize its molecular landscape and compare it with other complex sarcomas like UPS and MFS. Analyses revealed that TP53 (87%) and RB1 (51%) alterations are the most frequent in PLPS, and significantly more common than in UPS and MFS. MYC amplification was associated with significantly shorter overall survival, suggesting a prognostic role. Furthermore, correlations were established between gene alterations and morphological features, including CYSLTR2 alterations in cases with predominant pleomorphic lipoblasts. These findings highlight the molecular distinctiveness of PLPS and provide insights for future diagnostic and therapeutic strategies.
An integrated bioinformatics and functional study identified a metastasis-associated gene signature in osteosarcoma. Four key genes, AURKB, CDC20, KIF11, and TOP2A, were strongly associated with tumor progression and metastasis. Their high expression correlated with poor overall survival and demonstrated excellent diagnostic performance. A four-gene prognostic model showed strong predictive performance. Functional experiments confirmed their oncogenic role by promoting sarcoma cell proliferation and migration.
Cervical adenocarcinoma is linked to poorer survival and reduced response to immunotherapy. This study identified STK11 alterations, including mutations and focal deletions, as a distinct feature of HPV-positive cervical adenocarcinomas, affecting 23% of these tumors. These alterations are associated with a younger age at diagnosis, poorer overall survival, and inferior outcomes following immune checkpoint inhibitor therapy. Comprehensive genomic analyses revealed complex structural rearrangements involving chromosome 19p and STK11. These findings define an aggressive molecular subgroup and have implications for molecular classification and future targeted therapeutic strategies.
This study employed a multi-omics approach to decipher the metabolic-immune interplay in clear cell renal cell carcinoma (ccRCC). Researchers identified a five-gene metabolic signature (SUCLA2, ACAT1, PC, SUCLG1, HMGCS2) that predicts prognosis with superior accuracy compared to conventional clinical staging. This signature is associated with distinct tumor immune microenvironment states and enables patient stratification for personalized therapeutic strategies. In vitro experimental validations confirmed SUCLA2's role as a potential tumor suppressor, designating it as a promising target.
This study performed an integrated clinicogenomic analysis on over 7,000 colorectal cancer (CRC) samples to characterize its genetic drivers and metastatic tropisms. Researchers found that genomic evolutionary changes, such as clonal mutations and oncogenic mutant allelic imbalance, enhance the impact of recurrent oncogenic alterations. The study identified the relative timing of organ-specific metastasis, showing sequential progression in microsatellite stable CRC, with brain and adrenal metastases as late events. It also revealed clustering of metastatic sites and genomic events that modify metastasis risk, with WNT pathway activation being protective and RAS pathway activation increasing the risk of widespread metastasis. These data suggest that integrating information about primary tumor location and genomics can predict organ-specific metastasis risk.
This study introduces RGCNMDA, a novel multi-view learning framework designed to predict microRNA-disease associations, specifically addressing challenges posed by sparse data and "cold-start" entities. The model integrates global latent structures, local profiles, similarities, and biological pathway context through a relational graph. It employs a relational graph convolutional network for message passing and adaptive fusion of representations. RGCNMDA achieved robust performance on HMDD v4.0 and RNADisease v4.0 benchmarks, showing high AUCs, especially in cold-start scenarios. The research highlights the significant contribution of pathway context, particularly for less characterized diseases.
This study investigates cell-free DNA (cfDNA) profiles in the cerebrospinal fluid (CSF) of patients with lung cancer brain metastases. Using nanopore single-molecule sequencing, researchers analyzed fragmentation, methylation, and hydroxymethylation patterns of cfDNA. Significant differences were observed between cancer patients and healthy controls, including enriched mono-nucleosome levels and higher mono-/trinucleosome ratios in cancer patients. Distinct methylation and hydroxymethylation profiles were also identified, with a significantly lower degree of hydroxymethylation in cancer patients. These findings suggest that CSF cfDNA exhibits unique characteristics that could serve as biomarkers for non-small cell lung cancer brain metastases.
This prospective observational study investigated the prognostic value of circulating tumor DNA (ctDNA)-derived tumor fraction (TFx) and copy-number alterations (CNAs) in 78 patients with metastatic castration-resistant prostate cancer (mCRPC) receiving tandem [225Ac]Ac-/[177Lu]Lu-PSMA-617 radioligand therapy. Findings revealed that ctDNA-derived TFx serves as a dynamic biomarker for treatment response and progression risk, with a significantly increased risk of progression (HR 4.9) when modeled as a time-dependent variable. Furthermore, a high CNA burden was associated with shorter median overall survival (8.3 vs 13.8 months). The study identified recurrent CNAs affecting key tumor suppressor genes and highlighted the potential of these ctDNA-based biomarkers for clinical stratification and real-time monitoring.
This study investigated the transcriptomic profile and therapeutic opportunities in EGFR-mutant non-small cell lung adenocarcinoma (LUAD) that undergoes lineage transformation to small cell lung cancer (SCLC), a mechanism of acquired resistance to EGFR TKIs. By analyzing paired pre- and post-transformation samples from 32 patients, researchers characterized the immune microenvironment of transformed SCLC (T-SCLC). They found that T-SCLC exhibits an immunosuppressive microenvironment compared to EGFR-mutant LUAD, yet a more immunologically active one than primary SCLC. Patients with T-SCLC treated with first-line immunotherapy plus chemotherapy showed a median progression-free survival of 5.1 months and a median overall survival of 21.0 months. Immunotherapy exposure was associated with significantly longer overall survival, supporting the integration of immunotherapy in T-SCLC treatment.
This prospective translational study aims to identify integrated predictive and prognostic biomarkers (genomic, epigenomic, transcriptomic) from circulating tumor DNA (ctDNA) and tumor tissue for non-small cell lung cancer (NSCLC). The objective is to improve early detection, risk stratification, treatment selection and response prediction, and longitudinal monitoring. The project places a particular emphasis on reducing cancer inequities, especially within Māori communities in New Zealand. Methods include DNA sequencing, cfRRBS DNA methylation profiling, and RNA sequencing, with integrative bioinformatic analyses. Expected findings will support the development of sensitive, accessible, and equitable blood-based biomarkers for NSCLC detection and monitoring.
This study focuses on developing novel inhibitors targeting the RAS/RAF pathway, a significant challenge in oncology given the frequency of RAS mutations and acquired resistance to existing treatments. Researchers identified small molecules that covalently bind to a distinct site within the RAS-binding domain of RAF, leading to allosteric disruption of its conformation. This mechanism prevents RAS/RAF interaction and downstream signaling. These compounds demonstrated preclinical antitumor efficacy against various cancers with KRAS, NRAS, and HRAS mutations, as well as wild-type RAS-driven cancers. They also suppressed tumor growth in BRAFV600E melanoma with acquired BRAF inhibitor resistance by preventing RAS signal reactivation.
A retrospective study evaluated the efficacy and safety of first-line camrelizumab combined with chemotherapy (nab-paclitaxel and cisplatin) for advanced oesophageal squamous cell carcinoma (OSCC), compared to PD-1 monotherapy or chemotherapy alone. The study, involving 253 patients, demonstrated that the combination significantly improved progression-free survival (PFS) compared to the other arms. Although objective response and disease control rates did not differ, median overall survival was longest in the combination group (12.0 months). Adverse events were manageable, with higher rates of anemia and neutropenia in the combination group. An elevated neutrophil-to-lymphocyte ratio (NLR) was identified as an independent prognostic biomarker.
This study conducted deep multi-omic analyses on over 300 lung neuroendocrine tumours (NETs) and supra-carcinoids, integrating whole-genome sequencing, transcriptome profiling, and DNA methylation arrays. Researchers identified four strikingly different molecular groups, varying in patient characteristics, genomic and transcriptomic profiles, microenvironment, and morphology. A new molecular group, enriched for highly aggressive supra-carcinoids with an immune-rich microenvironment linked to tumor-macrophage crosstalk, was discovered. This morpho-molecular classification unifies previous proposals and offers potential implications for prognosis and therapeutic management, particularly through the identification of group-specific therapeutic targets such as DLL3, EGFR, FGFR, and TERT.
This study investigates the mechanisms underlying osteoblastic bone metastases in medullary thyroid cancer (MTC), which are associated with high mortality. Researchers found that activated RET protein in MTC cells stimulates the expression of osteoprotegerin (OPG), an inhibitor of bone resorption. This process leads to reduced osteoclast differentiation and promotes an osteoblastic phenotype. RET inhibition was shown to reduce tumor burden and bone lesions in mice. Furthermore, elevated circulating OPG levels in MTC patients with bone metastases correlate with poorer overall survival, suggesting OPG as a potential biomarker.
This study developed a robust prognostic model for lung adenocarcinoma (LUAD) by focusing on PD-1-related genes. Utilizing advanced bioinformatics analyses and in vitro validations, a nine-gene risk model (ADA2, CD53, HLA-DMA, ITGAX, MYO1F, NCKAP1L, PTPRC, RENBP, TNFAIP8L2) was constructed. This model demonstrated strong predictive performance, stratifying patients into high- and low-risk groups with distinct prognoses. High-risk patients exhibited a worse prognosis and lower immune infiltration, while drug sensitivity analyses identified specific candidate therapeutic agents for each group.
Zhonghua zhong liu za zhi [Chinese journal of oncology] ·
This study investigated the clinical value of dynamic monitoring of circulating tumor DNA (ctDNA) in 94 patients with advanced esophageal squamous cell carcinoma (ESCC) receiving sintilimab. The findings indicate that the ctDNA-based molecular tumor burden index (mTBI) serves as a sensitive, real-time biomarker for evaluating immunotherapy efficacy. Patients achieving ctDNA clearance or having a low mTBI after two treatment cycles demonstrated significantly delayed tumor progression and improved overall survival. Dynamic mTBI monitoring showed high consistency with concurrent imaging assessments, suggesting its utility for personalized treatment management.
This retrospective study investigated prognostic factors and outcomes of postmastectomy radiation therapy (PMRT) following neoadjuvant chemotherapy (NAC) and mastectomy for stage II-III breast cancer. It included 178 patients with a median follow-up of 64 months. The findings indicated that response to NAC, molecular subtype, and pathologic stage are independent predictors of treatment outcomes. PMRT did not offer significant benefits for patients with pathologically node-negative disease (ypN0) but significantly improved locoregional control, progression-free survival, and overall survival in those with residual lymph nodes (ypN1-3). Triple-negative subtype and non-response to NAC were identified as poor prognostic factors.
This paper introduces Biomni, a general-purpose artificial intelligence agent designed to autonomously execute and accelerate biomedical research. Biomni maps the biomedical action space by mining tools, databases, and protocols from thousands of publications, creating a unified agentic environment. Its architecture integrates large language model reasoning with retrieval-augmented planning and code-based execution, dynamically composing workflows without predefined templates. Systematic benchmarking shows strong generalization across diverse tasks, including causal gene prioritization, drug repurposing, and rare-disease diagnosis, without task-specific tuning. Biomni envisions augmenting human scientists and accelerating discovery.
This study investigates the interaction between mutated KRAS and AGO2 in colorectal cancer (CRC), a mechanism that impairs AGO2 function and promotes tumor progression. Researchers developed peptide-gold nanoparticles (Au-pep) to specifically disrupt this interaction, which restored microRNA expression, particularly let-7c-5p. This restoration led to the downregulation of key oncogenic drivers such as KRAS, c-Myc, and Bcl2, suppressing the growth of KRAS-mutated CRC cells and inducing apoptosis. Furthermore, this approach sensitized cells to existing treatments, including 5-fluorouracil and immunotherapy, by regulating ABCC5 and PD-L1.
This study characterized 33 RAD51C variants identified in breast cancer patients, assessing their ability to perform homologous recombination (HR). Researchers found one HR-deficient variant and five with intermediate HR proficiency. Although these variants exhibited only modestly attenuated HR function, they misregulate replication fork progression and accumulate single-strand DNA gaps. Cells expressing these variants, largely resistant to monotherapy, showed increased sensitivity to a combination of PARP1 inhibitors (Saruparib) and ATR inhibitors (Ceralasertib). These findings suggest that RAD51C variants with partial HR function can lead to profound defects in replicative damage repair, potentially resulting in distinct therapeutic responses.
This study investigated the expression of SWI/SNF chromatin remodeling complex genes and its subtypes (cBAF, PBAF, GBAF) across 33 tumor types, utilizing DNA and mRNA data from The Cancer Genome Atlas (approximately 11,000 samples). Researchers found that the expression of certain genes like ACTL6B, DPF1, DPF3, and SMARCD3 varied most among tumor types, with 10 out of 33 tumor types exhibiting distinct expression signatures. While SWI/SNF gene mutations resulted in only minor changes in their own expression, they were linked to downstream regulation of transcriptional programs associated with epigenetic regulation, particularly in colon adenocarcinoma, kidney renal clear cell carcinoma, stomach adenocarcinoma, and uterine corpus endometrial carcinoma. These findings could contribute to the development of biomarkers and targeted therapies.
This study characterized biallelic inactivation of the RB1 tumor suppressor gene (RB1-Bi) in hepatocellular carcinoma (HCC), finding it in 14.6% of patients and associating it with significantly reduced overall survival. A deep learning model (FR-MIL) was developed to accurately predict RB1-Bi status from routine histopathology slides. High-throughput drug screenings revealed that RB1-Bi HCC cells are particularly sensitive to inhibitors targeting mitotic regulators and PARP inhibitors. These synthetic lethal interactions were validated in vitro and in vivo, suggesting RB1-Bi is an actionable biomarker for precision medicine in HCC.
This study investigated the use of whole genome sequencing (WGS) to identify pathogenic variants previously undetected by routine genetic testing in patients with unexplained familial adenomatous polyposis (FAP). Researchers discovered two novel APC gene variants: a large deletion encompassing exon 10 and a deep intronic complex deletion-insertion variant. These variants were validated at the transcript level, confirming their impact on splicing and leading to truncated protein production. The findings underscore the importance of WGS and RNA-based analysis for comprehensive genetic diagnosis of FAP.
Accurate and efficient cell type annotation in single-cell RNA sequencing (scRNA-seq) data is a significant challenge, particularly due to batch effects and the need to handle large datasets. Existing deep learning methods often lack interpretability and robustness against these effects. To address these issues, the authors propose scKanFormer, a supervised framework built upon the Transformer architecture and Kolmogorov-Arnold Networks (KAN). This model integrates local and global attention mechanisms, along with biological information, to enhance the identification and interpretation of cellular features. scKanFormer demonstrated precise, robust, and transferable annotations, insensitive to batch effects, outperforming nine state-of-the-art methods on benchmark datasets.
This multicenter study developed a multimodal model (MuMo) to assess the prognosis of patients with locally advanced gastric cancer following neoadjuvant therapy. The model integrates radiomic features from preoperative CT images, post-neoadjuvant therapy transcriptomic sequencing data, and clinicopathological information. Machine learning algorithms identified independent radiomic and molecular risk factors. MuMo demonstrated superior performance compared to single-modality models and ypTNM staging in predicting overall and disease-free survival, suggesting the involvement of immune-inflammatory pathways in therapy efficacy.
This study identified and functionally validated lipid droplet (LD)-associated prognostic biomarkers in breast cancer through integrative multi-omics and machine learning. A prognostic model, built using the StepCox[both]+plsRcox algorithm, effectively stratified patients across molecular subtypes, outperforming traditional ER/PR/HER2 status. Key genes, SQLE and SOCS3, were identified with opposing roles: SQLE promoting adipogenesis and SOCS3 inhibiting it, influencing the tumor microenvironment and immune infiltration. Modulating these genes suppressed tumor growth and polarized macrophages towards an M1 phenotype, highlighting their potential as therapeutic targets.
This study introduces an extensive encyclopedia of human enhancer-gene regulatory interactions, encompassing over 92 million interactions across 1,458 biosamples and 369 cell types. Researchers developed ENCODE-rE2G, a state-of-the-art predictive model, by integrating various data sources including predictive models, chromatin states, three-dimensional contacts, and large-scale genetic perturbations from the ENCODE Consortium. The model was rigorously benchmarked using data from CRISPR perturbation experiments, expression quantitative trait loci (eQTL), and genome-wide association study (GWAS) variants. This resource offers valuable insights into enhancer networks and significantly improves the analysis of noncoding variants linked to complex diseases.
A study involving 5420 non-small cell lung cancer (NSCLC) cases investigated the prevalence and impact of coexisting driver genetic alterations. Coexisting mutations were found in 3.9% of tumors, more frequently in females, with KRAS and PIK3CA being the most common combination. Clonality analysis revealed KRAS and EGFR as initiating drivers, while PIK3CA, AKT1, and IDH1 were mostly secondary events. The presence of EGFR + KRAS mutations was significantly associated with worse overall survival in both early and advanced stages. These findings emphasize the importance of comprehensive genomic profiling for effective therapeutic management.
This study functionally characterized three BRCA2 variants of uncertain significance (VUS) identified in Korean breast cancer patients. Using BRCA2-knockout cell lines and various functional assays, researchers evaluated sensitivity to DNA-damaging agents and homologous recombination efficiency. The p.Asp1990Ala variant exhibited a pathogenic-like functional phenotype, characterized by increased sensitivity to cisplatin, ionizing radiation, and olaparib, along with impaired homologous recombination repair. In contrast, the p.Gly1224Asp and p.Cys1820Tyr variants were found to be functionally neutral. These findings highlight the heterogeneity of BRCA2 VUS and the importance of functional assays for their clinical interpretation.
This study describes a custom analysis method for detecting segmental copy number variants (CNVs) using an amplicon-based NGS panel (Oncomine Comprehensive Assay v3) for integrated glioma classification. The developed algorithm identifies critical biomarkers such as 1p/19q codeletion, +7/-10, EGFR amplification, and CDKN2A/B homozygous deletion. Validated across three cohorts, the approach demonstrated 100% concordance with FISH for all biomarkers in the validation cohort. Its clinical implementation significantly reduced required FISH studies by 90%, thereby shortening turnaround times and decreasing molecular testing costs.
An in vivo CRISPR screen identified BAP1 as a potent tumor suppressor in undifferentiated pleomorphic sarcomas (UPS). BAP1-deficient sarcomas exhibit a markedly immunosuppressive tumor microenvironment. Researchers found that PLK1 is required for the growth and survival of these sarcomas, with elevated expression in human UPS where BAP1 is reduced. Pharmacologic inhibition of PLK1 with volasertib significantly suppressed tumor growth in mouse models. Furthermore, combining PLK1 inhibition with anti-PD-1 therapy enhanced tumor control and improved survival.
This study performed integrated genomic, transcriptomic, and DNA methylation profiling on 76 desmoid tumors (DTs), uncovering frequent mutations in chromatin-remodeling genes, notably KMT2C, in addition to the known CTNNB1 and APC mutations. Transcriptomic analysis identified two robust molecular subtypes: immune-myogenic and mesenchymal-like. The immune-myogenic subtype exhibited high expression of myogenic markers and immune checkpoint genes, alongside enhancer hypomethylation. This subtype is associated with increased immune cell infiltration and predicts a higher responsiveness to immune checkpoint inhibitors.
Papadopoulou E, Stathoulopoulou M, Matiatou M, et al.
Cancer genomics & proteomics ·
This study investigated the associations between clinicopathological characteristics and genetic alterations in breast cancer, focusing on how clinical and pathological variables influence the actionability of Next Generation Sequencing (NGS) data. Among 227 breast cancer patients, comprehensive genomic profiling revealed that a vast majority (95.6%) harbored at least one oncogenic alteration. Subtype-specific patterns were identified, with an enrichment of PIK3CA and ESR1 alterations in HR+/HER2- tumors, and a predominance of TP53 in triple-negative breast cancer. The study highlights that NGS-based genomic profiling identifies clinically actionable alterations in over half of breast cancer patients, particularly in the HR+/HER2- subtype, and supports its routine implementation, especially for metastatic HER2-negative breast cancer.