Latest Research
All publications from the Cancer3.AI database, newest first.
An XGBoost-Based Multicenter Model for Predicting HBV-Related Hepatocellular Carcinoma: Development and Validation.
Lin Y, et al
Researchers developed and validated a machine learning model based on the XGBoost algorithm to predict hepatocellular carcinoma (HCC) caused by hepatitis B virus (HBV) infection, one of the leading causes of liver cancer globally. The multicenter retrospective study enrolled 3,568 participants across three major Chinese hospitals, identifying five key blood biomarkers—DCP, HBV DNA, ALT, AFP-L3%, and AFP—as the strongest predictors of HBV-related liver cancer using a combination of random forest, LASSO regression, and XGBoost methods. The XGBoost model demonstrated exceptional diagnostic accuracy, achieving area-under-the-curve (AUC) values of 0.985 in the training set, 0.978 in internal validation, and 0.942 in external validation, significantly outperforming four established clinical models including GALAD and ASAP. Beyond raw accuracy, the model showed superior individualized risk stratification, net clinical benefit, and calibration compared to existing tools, and an interactive web application was built to make the model readily accessible to clinicians. These findings could meaningfully improve early identification of high-risk HBV-infected patients, enabling timely surveillance and intervention that may ultimately improve survival outcomes for this vulnerable population.
Cancer medicine
Source →Early Detection of Gallbladder and Biliary Malignancy.
Gupta P, et al
A review article published in Radiologic Clinics of North America examines the significant challenges and emerging strategies surrounding the early detection of gallbladder cancer and cholangiocarcinoma, two highly aggressive biliary malignancies that carry poor prognoses largely because they are rarely caught at an early stage. These cancers are difficult to detect early because their initial symptoms are non-specific and frequently mimic benign biliary conditions, leading to diagnosis only when the disease has become advanced and curative options are limited. The article emphasizes that systematic risk stratification combined with multimodality imaging—including ultrasound, CT, and MRI—forms the cornerstone of any effective early detection strategy for these cancers. Emerging technologies such as artificial intelligence and deep learning algorithms are identified as promising tools capable of augmenting diagnostic accuracy when integrated into established imaging workflows. The authors advocate for a structured, multidisciplinary framework that unites risk-based surveillance, standardized high-quality imaging protocols, and innovative diagnostic technologies to address the persistent gap in early identification. Achieving earlier diagnosis holds the potential to substantially improve survival outcomes for patients with these otherwise devastating malignancies.
Radiologic clinics of North America
Source →Assessment of the Advancement of Alveolar Bone Loss in Canine Patients Receiving Definitive Radiotherapy for Sinonasal Tumors.
Paul-Stewart E, et al
A new retrospective study investigated whether radiation therapy for sinonasal (nose and sinus) tumors in dogs worsens the progressive loss of bone that supports teeth, known as alveolar bone loss, a well-recognized complication in human head and neck cancer patients. Researchers analyzed CT scans from dogs treated with definitive-intent radiotherapy at a veterinary teaching hospital between 2013 and 2022, measuring bone loss around the premolars and molars of both the upper jaw (maxilla, which received concentrated radiation) and the lower jaw (mandible, used as an internal control). While alveolar bone loss increased over time across all patients, the study found no statistically meaningful difference in the degree of bone loss between the irradiated maxilla and the non-irradiated mandible. This reassuring finding suggests that definitive radiotherapy for canine sinonasal tumors carries a low risk of significantly accelerating periodontal bone destruction, providing useful guidance for veterinary oncologists counseling owners about long-term side effects. As the first study of its kind in dogs, the authors call for continued long-term follow-up and expanded investigations into tumors located closer to the oral cavity to fully inform risk stratification and preventive dental care protocols.
Veterinary radiology & ultrasound : the official journal of the American College of Veterinary Radiology and the International Veterinary Radiology Association
Source →Disability-adjusted life-years from human papillomavirus-related oral cavity and pharynx cancers in US men, 2017-2021.
Patel HV, et al
A new study published in the journal Cancer quantified the health burden of human papillomavirus (HPV)-associated oropharyngeal cancers — specifically cancers of the tonsils and base of the tongue — in US men between 2017 and 2021, using disability-adjusted life-years (DALYs), a metric that combines years of life lost to premature death with years spent living with illness-related disability. Researchers drew on data from the National Cancer Institute's SEER 17 database and applied population-attributable fractions to isolate HPV's contribution to these cancers, finding that average annual DALYs were highest for tonsillar cancer at approximately 58,979 per year, followed by base of tongue cancer at approximately 40,440 per year. Overall DALY trends were encouraging, showing statistically significant decreases of 2.44% per year for tonsillar cancer and 4.28% per year for base of tongue cancer, likely reflecting advances in treatment and earlier diagnosis. However, a concerning exception emerged among men aged 65 and older, for whom DALYs increased across both cancer sites, a group that was never eligible for HPV vaccination when it was first introduced. These findings underscore the critical importance of gender-neutral HPV vaccination policies and targeted catch-up vaccination programs for older men who remain at elevated risk, as HPV-related oropharyngeal cancer in men has now surpassed cervical cancer in women as the most common HPV-linked malignancy.
Cancer
Source →Gaps in Guideline Adherence for Primary Sclerosing Cholangitis in North America-A 5-Year Analysis.
Martins A, et al
A new large-scale study examined how well North American physicians follow established clinical guidelines for primary sclerosing cholangitis (PSC), a rare progressive liver disease that significantly raises the risk of liver and colon cancers. Researchers analyzed data from 1,300 patients enrolled across 19 centers between 2018 and 2024, finding that adherence to key recommended practices was consistently suboptimal. Only about one-third of PSC patients with inflammatory bowel disease received the recommended annual colonoscopy for colorectal cancer surveillance, and imaging for liver cancer detection dropped from 74% in year one to just 51% by year four. Socioeconomic factors such as lower education and household income were linked to poorer guideline adherence, while having cirrhosis was the strongest predictor of completing surveillance imaging. These findings highlight critical gaps in the real-world management of PSC and underscore the urgent need to standardize care pathways and reduce socioeconomic barriers to improve patient outcomes.
Liver international : official journal of the International Association for the Study of the Liver
Source →