Find the papers that actually matter

Microscopy-based cancer cell classification traditionally relies on cell-based morphological features, while subcellular organelle organization remains underutilized. Existing machine learning methods often require manual preprocessing and handcrafted feature extraction, limiting scalability and introducing user bias. This study proposes an automated, interpretable, and organelle-focused deep learning framework for classifying breast cancer cell lines from high-resolution fluorescence microscopy images. We developed an end-to-end framework that incorporates patch-based sampling, sparsity filtering, and a channel-wise intermediate fusion strategy to independently extract and integrate organelle-specific features. Model interpretability was assessed using Grad-CAM visualizations and single-organelle classifier analyses. The framework was evaluated on fluorescence microscopy images from six breast cancer cell lines using 5-fold cross-validation. The proposed framework achieved a classification accuracy of 97.1 ± 1.1 %, performing comparably to or exceeding conventional handcrafted feature-based approaches while eliminating the need for manual segmentation and 3D rendering steps. Interpretability and classifier analyses revealed inter-organelle dependencies and mitochondria as the most informative contributors to classification decisions. Organelle morphology and spatial organization provide strong discriminative signals for cancer cell classification. The proposed framework offers a scalable, automated, and interpretable deep learning solution that advances microscopy-based phenotyping and supports broader applications in computational pathology and cellular informatics.

Nasopharyngeal carcinoma (NPC) poses significant diagnostic challenges due to the anatomical complexity of the nasopharynx and reliance on endoscopic visual interpretation, often leading to delayed detection and unnecessary biopsies. Although artificial intelligence (AI) algorithms have shown promise in enhancing endoscopic cancer diagnosis, their real-world impact on clinician diagnostic performance remains insufficiently characterized. In this prospective, multi-reader study, 47 clinicians, including experts, residents, and trainees, interpreted 200 nasoendoscopic images from 100 patients with histopathologically confirmed NPC or benign lesions. Each participant completed two diagnostic sessions: an unassisted evaluation and an AI-assisted assessment using a real-time lesion-annotating model (NPC-SDNet), with a 4-week washout period between sessions. Without AI support, the overall diagnostic accuracy was 73.6% (95% CI: 70.1-77.0%), with a sensitivity of 76.1% (95% CI: 70.4-81.4%) and a specificity of 69.9% (95% CI: 63.8-76.0%). AI assistance significantly improved accuracy to 85.6% (95% CI: 83.1-87.6%, p < 0.001), sensitivity to 90.1% (95% CI: 86.6- 92.9%, p < 0.001), and specificity to 79.1% (95% CI: 75.6-82.7%, p < 0.001). Subgroup analysis revealed the greatest improvements among trainees (64.8% vs 83.5%, p < 0.001) and residents (77.2% vs 84.9%, p = 0.003). Moreover, AI integration substantially reduced median image interpretation time from 1411.7 to 818.5 s (p < 0.001). AI-assisted nasoendoscopic evaluation significantly enhances diagnostic accuracy, efficiency and interobserver consistency, particularly among less-experienced clinicians. These findings support the clinical integration of real-time AI tools to augment NPC recognition and standardize diagnostic performance across varying expertise levels.

Inadequate adherence to colorectal cancer screening reduces individual and population level health benefits. Blood-based tests offer a new modality that may help patients overcome barriers, but there are concerns about the impact of patients switching from existing guideline-recommended screening modalities. This study estimates the population health outcomes and cost-effectiveness of offering blood-based testing using a validated individual-level simulation model based on patient preference evidence from randomized controlled trials. In this study, a validated discrete-event simulation model was used to evaluate the performance of different combinations of colorectal cancer screening strategy preferences per 10,000 screened individuals beginning at age 45. Preferences for screening options were informed by randomized controlled trials of patients with and without the option of blood-based testing. Adherence to initial patient preferences over a simulated lifetime was modeled as: (1) assumed 100% adherence and (2) longitudinal using a calibrated model. Simulated outcomes included clinical outcomes and cost-effectiveness from a healthcare sector perspective. A strategy was deemed cost-effective at a willingness-to-pay threshold of $100,000 per quality-adjusted life-year gained. The introduction of blood-based testing to an unscreened population with evidence from randomized controlled trials is projected to increase colorectal cancer deaths averted by 35% to 116% and from 68% to 247% relative to no screening, for stated preference and revealed preference scenarios, respectively. These outcomes are cost-effective, with incremental cost-effectiveness ratios ranging from $63,994 to $85,497 and from $30,464 to $54,764 across stated preference and revealed preference scenarios, respectively. Given limited data, natural history and real-world longitudinal adherence to screening are based on evidence-informed assumptions. Using a simulation model to extrapolate data from two recent trials, we demonstrate that the introduction of blood-based tests has the potential to lead to cost-effective increases in the number of CRC deaths averted among the unscreened population.

Kidney transplantation (KT) is the most effective treatment for end-stage renal disease. Hypothermic machine perfusion (HMP) can improve renal energy metabolism and reduce ischemia-reperfusion injury compared with static cold storage. This study aimed to evaluate the association between HMP parameters and graft function in deceased donor kidney transplantation (DDKT) and to develop a predictive model for early risk stratification. A retrospective analysis was conducted on 2,041 DDKT recipients from 1 January 2015 to 30 June 2023. The primary outcome, delayed graft function (DGF), was defined as the need for at least one dialysis session within the first week after transplantation. Consensus clustering (CC) and restricted cubic spline (RCS) analysis were used to evaluate the associations between clinical data, HMP parameters, and graft function. Feature selection was performed using Lasso-penalized logistic regression (LR), and multivariable LR was used to construct the predictive model. The model's performance was assessed using the area under the curve (AUC), calibration curves, and decision curve analysis (DCA). Among the DDKT recipients, 12.9% developed DGF. HMP parameters varied significantly between the two groups, with DGF recipients showing distinct patterns in perfusion resistance, flux, and pressure. CC identified two recipient clusters with distinct DGF risk profiles, graft function, and donor characteristics. Non-linear relationships were identified between HMP parameters and DGF risk, with thresholds for initial resistance, terminal resistance, and terminal flux. The predictive model integrating six variables achieved an AUC of 0.78 (95% CI: 0.76-0.82) in the test set. Calibration and DCA confirmed good reliability and net clinical benefit. Non-linear relationships between HMP parameters and DGF underscore graft perfusion complexity. The proposed model demonstrated robust internal performance and may support early post-transplant risk stratification. External validation in independent cohorts is warranted to confirm generalizability and clinical applicability.

This study aimed to identify neutrophil extracellular trap-related genes (NET-RGs), explore their prognostic significance, and predict drug interactions in peripheral T-cell lymphoma (PTCL). Differentially expressed NET-RGs (DE-NRGs) between PTCL and normal tissues were screened. Functional enrichment analysis was conducted. Bioinformatics and machine learning were used to identify hub genes and assess their diagnostic value. Univariate and multivariate analyses were used to evaluate prognostic roles. Correlation and immune infiltration analyses were performed to explore relationships with the tumor microenvironment (TME). Clinical data were collected from PTCL patients who received potential agents (lenalidomide) as first-line treatment. A total of 31 DE-NRGs were identified (18 upregulated and 13 downregulated), enriched in inflammatory response, extracellular matrix organization, and infection involvement. Four hub genes (AKT2, MAPK14, IRF1, and TNF) were identified as effective PTCL diagnostic markers. Higher AKT2/MAPK14 expression correlated with poorer overall survival (OS), while elevated TNF expression associated with better OS; AKT2 and TNF independently predicted OS. These genes were implicated in modulating TME remodeling. Potential therapeutic agents (e.g. capivasertib, lenalidomide) were predicted, and lenalidomide may represent a feasible initial treatment option for PTCL, with an objective response rate (ORR) of 40.0% and a maximum survival duration exceeding 50 months. NET-RGs play crucial roles in diagnosis, prognosis, and TME regulation, and lenalidomide, a putative TNF-targeting agent, may represent a feasible initial treatment option in PTCL.

Diethyl phthalate (DEP), a common plasticiser and endocrine disruptor, has been linked to cancer, but its role in osteosarcoma (OS) remains unclear. This study integrated network toxicology, transcriptomics, protein-protein interaction (PPI) analysis, machine learning, molecular docking, molecular dynamics (MD), single-cell RNA sequencing (scRNA-seq), and external validation to investigate DEP-related mechanisms in OS. We identified 45 DEP-responsive genes enriched in extracellular matrix-related pathways. PPI network analysis revealed 11 hub genes, of which LASSO, SVM-RFE, and Boruta algorithms consistently prioritised P4HA2, COL18A1, and COL10A1. Docking and MD simulations supported stable binding of DEP to P4HA2 and COL18A1 via hydrogen bonds and hydrophobic interactions. scRNA-seq demonstrated celltype-specific expression of these genes. Validation cohorts confirmed their upregulation in OS, with AUC values up to 0.950. These findings suggest that DEP may promote OS progression by targeting extracellular matrix remodelling, offering new diagnostic biomarkers and hypothesis-generating evidence for environmental osteocarcinogenesis.

Docetaxel (DTX) resistance is the main cause of treatment failure in castration-resistant prostate cancer (CRPC). Pyrroline-5-carboxylic acid reductase 1 (PYCR1) is an enzyme involved in proline metabolism. It is highly expressed in various cancers and promotes malignant progression, yet its role in DTX resistance in prostate cancer remains unclear. In this study, bioinformatics analyses and in vitro/vivo experiments demonstrated that interfering with PYCR1 expression modulates the sensitivity of prostate cancer cells to DTX. Subsequently, via structure-based virtual screening, molecular dynamics simulations, and cellular thermal shift assay (CETSA), emodin-an anthraquinone compound-was identified as a PYCR1-targeting agent. Collectively, these findings suggest that PYCR1 may serve as a key target mediating DTX resistance in prostate cancer, and the emodin-DTX combination provides a promising potential clinical strategy to overcome such resistance. Finally, its functions and safety were also verified through in vitro experiments.

Although FGFR2 is a well-validated oncogenic target, no selective FGFR2 inhibitors have been approved for clinical use. In this study, we report the discovery of 2H-pyrazolo[3,4-d]pyrimidin-4-amine derivative as novel, irreversible FGFR2 inhibitors. The optimal compound, PLW559, potently inhibited FGFR2 with an IC50 value of 13.59 nM and demonstrated exceptional selectivity over FGFR1, FGFR3, and FGFR4. Covalent binding to the target was confirmed by mass spectrometry. In cellular models, PLW559 exhibited potent and selective antiproliferative effects against FGFR2-driven cancer cells, effectively suppressed downstream FGFR2 signalling and induced cancer cell apoptosis. Notably, it showed minimal activity in non-FGFR2-dependent cells. This work presents a new class of selective FGFR2 inhibitors based on a novel scaffold, offering promising lead compounds for the development of FGFR2-target therapies.

ABCB1-mediated drug efflux is a key determinant of multidrug resistance (MDR) in cancer. To overcome this mechanism, a series of thiol-substituted aminocoumarin-derived, coumarin-containing triazolo[1,5-a]pyrimidine derivatives (5a-5s) was synthesised, and compound 5r (NYH-707) was identified as the most potent ABCB1 inhibitor. NYH-707 markedly restored paclitaxel sensitivity in SW620/Ad300 MDR cells, reducing the IC50 from 4.55 ± 0.73 µM to 0.011 ± 0.002 µM (reversal factor = 413.6). Molecular docking predicted strong binding (-9.7 kcal/mol) through hydrogen bonding with LYS-826 and SER-880 and π-π stacking with PHE-994. CETSA confirmed direct ABCB1 engagement, while drug-accumulation assays demonstrated inhibition of ABCB1-mediated efflux. In vivo, co-administration of NYH-707 and paclitaxel significantly suppressed SW620/Ad300 xenograft growth without detectable systemic toxicity. These findings indicate that NYH-707 acts as a potent and selective ABCB1 modulator capable of reversing MDR likely by modulating ABCB1 conformational dynamics, thereby enhancing chemotherapeutic efficacy in resistant tumours.

FMS-like tyrosine kinase 3 (FLT3/CD135) regulates haematopoiesis and is frequently mutated as FLT3-internal tandem duplication (FLT3-ITD) in acute myeloid leukaemia (AML), associated with poor prognosis. Although FLT3 inhibitors show clinical benefits, resistance remains a challenge. This study hypothesises that antibody-drug conjugate (ADC) efficacy depends on distinct FLT3 trafficking mechanisms in FLT3-wt and FLT3-ITD cells. Confocal imaging showed that in THP-1 (FLT3-wt) cells, FLT3 mAb trafficked to lysosomes, while in MV4-11 (FLT3-ITD) cells, it accumulated in the Golgi. To evaluate the impact of this trafficking difference, we synthesised an anti-FLT3 mAb-MMAE, linked via a Val-Cit-PAB linker at the Fc N-glycan, which exhibited lower cytotoxicity in MV4-11 than THP-1 cells, indicating that the impaired lysosomal trafficking of FLT3-ITD limits drug release and reduces ADC potency. These findings highlight that effective lysosomal targeting is essential for ADC activity and suggest that optimising linker design or restoring lysosome trafficking may enhance FLT3-targeted ADC in AML.