Latest Research
All publications from the Cancer3.AI database, newest first.
Concordance and Acceptability of Self- Vs. Clinician-Collected Anorectal Swabs for HPV Genotyping in Gay, Bisexual and Other Men Who Have Sex With Men in Metropolitan Barcelona.
Ferrera L, et al
A multicenter cross-sectional study conducted in metropolitan Barcelona evaluated whether gay, bisexual, and other men who have sex with men could accurately collect their own anal swabs for human papillomavirus (HPV) genotyping, compared to samples taken by a clinician. Among 151 participants—over half of whom were living with HIV—self-collected swabs detected HPV in 89.9% of cases versus 92.6% for clinician-collected swabs, with an overall agreement of 96.0% and a strong Cohen's kappa of 0.75. Sensitivity of self-sampling reached 96.4% and specificity 90.9%, with comparable accuracy for high-risk HPV types including HPV16, which are most closely linked to anal cancer. The majority of participants found the self-sampling procedure acceptable and feasible, suggesting it could overcome existing barriers to screening access. These findings support anal self-sampling as a reliable, patient-centered strategy to improve early detection of HPV-related precursor lesions and reduce the burden of squamous cell carcinoma of the anus in high-risk populations.
Journal of medical virology
Source →DNA Virus Detection in Olfactory Neuroblastomas Using Targeted Enrichment NGS.
Jauhiainen MK, et al
Researchers conducted the first comprehensive viral investigation of olfactory neuroblastoma (ONB), a rare malignant tumor arising in the olfactory region of the nasal cavity with an unknown cause. Using targeted next-generation sequencing capable of detecting 41 different DNA viruses — including papillomaviruses, herpesviruses, polyomaviruses, and parvoviruses — alongside quantitative PCR for herpesviruses, the team analyzed tissue samples from 12 ONB patients. The study found no significant viral presence in ONB tumors, with the notable exception of low-risk HPV6 detected in one sample, suggesting that DNA viruses do not play a major role in causing this cancer. Human endogenous retroviruses were detected in all samples, confirming the technical reliability of the sequencing process. The detection of low-risk HPV6 is clinically relevant because this virus has previously been linked to an increased risk of cancer progression in benign sinonasal papillomas arising in the same anatomical region. These findings largely rule out a viral etiology for ONB but underscore the need for further research into the potential role of HPV in sinonasal tumor development.
Neuropathology : official journal of the Japanese Society of Neuropathology
Source →Radiomics in Otolaryngology-Head and Neck Surgery: A Review.
Bourdillon AT, et al
A comprehensive review published in JAMA Otolaryngology–Head & Neck Surgery examined the current state of radiomics—a technology that extracts hundreds of quantifiable features from medical images to uncover hidden patterns linked to disease biology and patient outcomes—across the full spectrum of ear, nose, and throat (otolaryngology) medicine. Researchers found that radiomics has been applied widely to cancers of the upper airway, larynx, sinuses, and skull base, enabling detection of clinically hidden features such as lymph node spread and extranodal extension that directly influence treatment decisions. Advanced approaches including dosiomics (analyzing radiation dose distributions) and delta-radiomics (tracking imaging changes over time) showed particular promise for predicting how tumors respond to radiotherapy and for anticipating treatment-related side effects. Hybrid models that combine radiomics data with conventional clinical information consistently outperformed either source of data used alone. Despite this broad potential, the review identified key barriers to clinical adoption, including inconsistent imaging protocols, labor-intensive image segmentation, small study populations, and a lack of validation across multiple institutions. The authors concluded that standardizing imaging workflows and conducting large multi-institutional trials are essential next steps before radiomics tools can be routinely integrated into clinical practice to improve diagnosis, prognosis, and personalized treatment planning for head and neck patients.
JAMA otolaryngology-- head & neck surgery
Source →Explainable machine learning to predict muscle loss during radiotherapy for oral cavity cancer.
Lin J, et al
Researchers developed and validated an explainable machine learning model to predict muscle loss in patients with oral cavity cancer undergoing surgery followed by adjuvant radiotherapy, a complication linked to poor survival. The study analyzed data from 1,024 patients treated between 2010 and 2021, using CT scans at the neck level to measure changes in muscle mass before and after radiotherapy. Among three tested models — Random Forest, XGBoost, and CatBoost — the Random Forest model achieved the highest accuracy in an independent external validation cohort, with an area under the curve of 0.913, indicating excellent predictive performance. Key risk factors identified included nutritional status before radiotherapy, radiation doses delivered to swallowing muscles and the supraglottic larynx, and the use of chemotherapy, with a nonlinear relationship observed between radiation dose to swallowing structures and muscle loss. The model also provided patient-level explanations using the SHAP method, allowing clinicians to understand which factors drove the prediction for each individual. These findings could enable personalized radiotherapy planning and early nutritional or rehabilitation interventions to reduce muscle loss and improve long-term outcomes for oral cavity cancer patients.
Radiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology
Source →Identification and Localization of Functioning Gastroenteropancreatic Neuroendocrine Tumors Using [18F]F-NOTA-octreotide PET/CT.
Lin L, et al
Researchers investigated the diagnostic utility of [18F]F-NOTA-octreotide PET/CT, a molecular imaging technique, for detecting and localizing functioning gastroenteropancreatic neuroendocrine tumors (F-GEP-NETs) — rare tumors that overproduce hormones and cause distinct clinical syndromes. Two illustrative cases were presented: a gastrinoma and a VIPoma, both of which went undetected or produced ambiguous results on standard contrast-enhanced computed tomography (CE-CT). In both instances, [18F]F-NOTA-octreotide PET/CT successfully identified and precisely localized the primary tumors by exploiting the overexpression of somatostatin receptors on neuroendocrine tumor cell surfaces, providing superior functional imaging over conventional anatomical methods. Conventional imaging techniques such as CT and MRI frequently fail in this setting because F-GEP-NETs are often very small and visually heterogeneous, leading to false-negative results that delay diagnosis and treatment. These findings demonstrate that [18F]F-NOTA-octreotide PET/CT holds significant clinical value as a next-step diagnostic tool when standard imaging is inconclusive, enabling clinicians to formulate more accurate surgical strategies and personalized management plans for affected patients.
Journal of visualized experiments : JoVE
Source →