Latest Research
All publications from the Cancer3.AI database, newest first.
Rectal Prolapse as a Rare Presentation of Colorectal Cancer: Case Series and Clinical Perspectives.
Nugraha P, et al
A case series published in Clinical and Experimental Gastroenterology describes three patients in whom rectal prolapse — the protrusion of the rectum through the anal opening — served as an unusual and initial sign of underlying colorectal cancer. The patients, aged 40, 79, and 79, presented with rectal prolapse either in an outpatient setting or as a surgical emergency, and all underwent tailored surgical resections followed by histological examination that confirmed adenocarcinoma in each case, staged from pT1 to pT2 without lymph node involvement. Notably, one patient had a cancer of the descending colon rather than the rectum itself, demonstrating that the association between prolapse and malignancy can involve unexpected anatomical sites. The authors stress that preoperative imaging is critical in all cases of rectal prolapse — especially in patients who lack typical predisposing factors such as advanced age or pelvic floor weakness — to detect occult tumors before surgery. This report serves as an important reminder to clinicians that colorectal cancer should be actively excluded as a potential cause of rectal prolapse, since early identification of malignancy directly shapes the surgical strategy and reduces the risk of recurrence and complications.
Clinical and experimental gastroenterology
Source →Artificial intelligence reshaping the paradigm of hematologic malignancy diagnosis and treatment: From static assessment to dynamic precision management.
Li Z, et al
A comprehensive review published in Annals of Hematology examines how artificial intelligence is transforming the diagnosis and treatment of hematologic malignancies, including leukemia, lymphoma, multiple myeloma, and myelodysplastic syndromes. Unlike earlier reviews confined to single disease types, this paper synthesizes AI applications across morphology analysis, medical imaging, flow cytometry, and multimodal data fusion into a unified conceptual framework. The authors propose an AI-driven dynamic diagnosis and treatment system integrated with electronic health records, designed to support continuous risk stratification and longitudinal patient monitoring in routine clinical workflows. The review finds that AI holds strong promise for improving diagnostic consistency, refining risk stratification, and enabling more personalized treatment decisions across the full spectrum of blood cancers. However, the authors identify significant barriers to clinical adoption, including limited and siloed training data, poor cross-institutional model generalization, and unresolved concerns around interpretability and ethical oversight. To move the field forward, the authors recommend multicenter prospective validation studies adhering to international standards such as TRIPOD+AI, alongside robust frameworks for data privacy and algorithmic accountability that could ultimately support more dynamic and individualized blood cancer management.
Annals of hematology
Source →Utility of Iron Staining in Bone Marrow Aspirates in the Era of Next-Generation Sequencing: A Retrospective Single-Centre Study.
Ganeshalingam V, et al
Researchers at a regional Australian hospital conducted a retrospective study to determine whether routine bone marrow iron staining retains meaningful diagnostic value in the modern era of next-generation sequencing (NGS). The study analyzed 101 consecutive adult bone marrow examinations performed between January 2023 and June 2024, covering both myeloid and lymphoid blood cancers. Iron staining had no diagnostic impact in 94% of cases, and in the remaining 6% it was only supportive and non-essential — meaning omitting the test would not have changed any final diagnosis. Ring sideroblasts, a morphological feature historically central to classifying certain blood disorders such as myelodysplastic syndromes, were found in seven myeloid cases but did not independently determine diagnosis, with molecular testing for SF3B1 gene mutations providing more definitive information. Furthermore, marrow iron grading correlated poorly with blood ferritin levels, limiting its usefulness as a surrogate marker for whole-body iron status. The authors conclude that a targeted rather than routine approach to bone marrow iron staining is warranted — particularly where NGS is available — which could streamline diagnostic workflows and reduce unnecessary testing.
International journal of laboratory hematology
Source →A 2026 update on myelodysplastic neoplasms: current state, challenges and future directions.
Bewersdorf JP, et al
A comprehensive 2026 review published in Nature Reviews Clinical Oncology examines the current state of myelodysplastic neoplasms (MDS), a heterogeneous group of blood cancers that primarily affect older adults, with a median diagnosis age of 76 years in the United States. MDS is defined by ineffective blood cell production, low blood counts, and a variable risk of transforming into acute myeloid leukemia, yet despite two decades of scientific progress, only a handful of disease-modifying therapies have been approved, and allogeneic hematopoietic stem cell transplantation remains the sole potentially curative option. The authors explain that slow therapeutic progress stems from MDS's complex biology, involving interactions between somatic and inherited genetic mutations, immune system dysfunction, and chronic inflammation in the bone marrow microenvironment. The review also addresses how recent updates to diagnostic tools, disease classification systems, and prognostic models, while scientifically valuable, have created new challenges for clinical trial design and epidemiological tracking. Ultimately, the authors outline future research priorities aimed at developing novel and effective treatments, underscoring that improving outcomes for the growing population of elderly MDS patients remains an urgent unmet medical need.
Nature reviews. Clinical oncology
Source →A deep learning-driven automated treatment planning framework for cervical cancer patients treated with volumetric modulated arc therapy.
Ning B, et al
Researchers developed a deep learning-based end-to-end automated treatment planning framework for cervical cancer patients receiving volumetric modulated arc therapy (VMAT), a highly precise form of radiation treatment. The system was trained and validated on 458 patient treatment plans and employs a two-stage cascaded neural network: the first stage predicts an initial dose distribution from CT scans and anatomical structure masks, while the second stage refines that prediction using beam-geometry information and a composite loss function. Compared to manually created clinical plans, the automated system maintained equivalent tumor coverage while reducing radiation dose to healthy surrounding organs — including the bladder, rectum, small intestine, and spinal cord — by between 2% and 35%, with all differences being statistically significant. The system also achieved a marginally higher gamma passing rate (98.1% versus 97.9%), confirming that the automatically generated plans can be accurately delivered by radiotherapy machines. These findings suggest the framework could substantially streamline radiotherapy workflows, improve plan consistency across patients, and reduce the planning burden on radiation oncologists without compromising treatment quality.
Radiation oncology (London, England)
Source →