Latest Research
All publications from the Cancer3.AI database, newest first.
RHBDL2 drives lipid metabolic reprogramming in osteosarcoma via USP3-mediated deubiquitination of PPT1.
Fan L, et al
Osteosarcoma is an aggressive bone cancer with notoriously poor outcomes, and researchers have now uncovered a previously unknown molecular signaling axis that drives its abnormal fat metabolism. The study found that a protein called RHBDL2 is significantly overexpressed in osteosarcoma tissue and functions as a non-enzymatic molecular scaffold that stabilizes the deubiquitinase USP3, an interaction anchored by a key hydrophobic residue (Val245) and independent of RHBDL2's protease activity. Stabilized USP3 in turn prevents the degradation of PPT1, a metabolic regulator that fuels tumor growth by activating FASN-driven de novo lipid synthesis, collectively promoting cancer cell proliferation, migration, epithelial-mesenchymal transition, and resistance to cell death. Critically, the researchers identified Epigallocatechin gallate (EGCG), a natural compound abundant in green tea, as a potent inhibitor that competitively disrupts the RHBDL2-USP3 protein interface, thereby shutting down the downstream lipogenic program and suppressing tumor growth and bone destruction in animal models. These findings define the RHBDL2-USP3 structural interface as a promising and druggable therapeutic vulnerability, offering a new mechanistic rationale for targeting lipid metabolic reprogramming in osteosarcoma patients.
Cell death & disease
Source →Hybrid diagnostic framework for bone cancer detection using deep learning and radiomics analysis.
Ramamoorthy R, et al
Researchers have developed TriMedNet, a novel hybrid deep learning framework designed to improve the early and accurate detection of bone cancer by simultaneously analyzing multiple types of patient data. The system integrates MRI scan images, unstructured clinical text notes, and structured numerical patient metrics such as blood pressure and glucose levels through three specialized processing branches: a Convolutional Neural Network for image analysis, a BERT transformer model for text interpretation, and dense neural layers for numerical data. Features extracted from each branch are fused and passed to a final classification layer that produces a tumor diagnosis. When evaluated on a publicly available dataset that included biopsy and blood test results, TriMedNet achieved an accuracy of 98.5%, a precision of 97.6%, and a recall of 98.2%, outperforming single-modality diagnostic approaches. These findings demonstrate that combining imaging, textual, and numerical clinical information significantly enhances diagnostic performance, offering clinicians a powerful decision-support tool for earlier and more reliable bone cancer diagnosis.
Scientific reports
Source →LncRNA MCM3AP-AS1 is a potential prognostic marker for osteosarcoma and promotes its development by targeting miR-205-5p.
Liu Y, et al
Researchers investigated the role of a long non-coding RNA (lncRNA) called MCM3AP-AS1 in osteosarcoma, a malignant bone tumor that most commonly affects children and adolescents and carries a poor prognosis in advanced cases. The study found that MCM3AP-AS1 is differentially expressed in osteosarcoma tissue compared to normal bone, and its expression levels correlate with patient survival outcomes, positioning it as a potential prognostic biomarker. Mechanistically, MCM3AP-AS1 was shown to promote tumor growth and progression by acting as a molecular sponge that sequesters and suppresses miR-205-5p, a microRNA known to inhibit cancer cell proliferation. By silencing miR-205-5p, MCM3AP-AS1 effectively releases downstream oncogenic targets, accelerating osteosarcoma development. These findings illuminate a previously underappreciated regulatory axis in bone cancer biology and suggest that MCM3AP-AS1 could serve both as a diagnostic prognostic tool and as a novel therapeutic target for improving outcomes in osteosarcoma patients.
Discover oncology
Source →Patients with myelodysplastic/myeloproliferative neoplasms have higher rates of relapse following allogeneic hematopoietic cell transplantation compared to those with MDS: A Brazilian SBTMO/CIBMTR registry analysis.
Barroso Duarte F, et al
This registry-based study from Brazil compared outcomes of allogeneic hematopoietic cell transplantation (allo-HCT) between patients diagnosed with myelodysplastic/myeloproliferative neoplasms (MDS/MPN) and those with classic myelodysplastic syndromes (MDS), using data from the SBTMO/CIBMTR collaborative database. The investigators found that patients with MDS/MPN experienced significantly higher rates of disease relapse following allo-HCT than their MDS counterparts, indicating that the overlap syndrome carries a more aggressive post-transplant disease course. This distinction is clinically important because MDS/MPN is a heterogeneous group of blood cancers that combines features of both dysplastic and proliferative disorders, making it biologically distinct and potentially less responsive to transplant-induced graft-versus-disease effects. The findings suggest that current transplant strategies optimized for MDS may not be sufficient for MDS/MPN patients, and that disease-specific approaches to maintenance, conditioning, or post-transplant surveillance should be explored. This real-world evidence from a Latin American cohort also broadens the global understanding of transplant outcomes beyond predominantly European and North American registries, offering important data for clinicians treating diverse patient populations.
Bone marrow transplantation
Source →Longitudinal localization of leukaemic stem cells between the metaphysis and central marrow governs their behaviour.
Wang C, et al
This study investigated how the physical location of leukemic stem cells (LSCs) within the bone marrow controls their behavior and resistance to therapy in acute myeloid leukemia (AML). Researchers discovered that LSCs preferentially reside in the metaphysis — the bone growth region — where they remain dormant and highly protected, whereas their displacement to the central marrow reduces their stem-like properties and aggressiveness. A key molecular pathway called the CXCL12-DPP4-GPC3 axis was identified, in which the enzyme DPP4 inactivates the signaling molecule CXCL12, while a stromal protein called GPC3 restrains DPP4 activity, collectively maintaining chemical gradients that anchor LSCs to their protective niche at multiple spatial scales. By pharmacologically targeting DPP4 in AML models, scientists disrupted these gradients simultaneously at the whole-body level, within the bone marrow architecture, and at the microscale around supportive stromal cells, dislodging LSCs from the metaphysis and rendering them more vulnerable to chemotherapy. These findings demonstrate that the spatial address of cancer stem cells is not incidental but actively governs their therapeutic resistance, pointing to niche disruption as a promising strategy to overcome relapse in AML patients.
Nature cell biology
Source →