Latest Research
All publications from the Cancer3.AI database, newest first.
The prognostic value of the lung immune prognostic index in patients with urological cancers: a systematic review and meta-analysis.
Lyu Z, et al
A new systematic review and meta-analysis published in Frontiers in Immunology evaluated whether the Lung Immune Prognostic Index (LIPI) — a simple blood-based score combining a neutrophil-to-lymphocyte ratio and lactate dehydrogenase levels — can predict survival outcomes in patients with urological cancers, including kidney, bladder, and prostate cancers. Researchers pooled data from 13 studies encompassing 19 independent patient cohorts and over 5,300 patients, using rigorous statistical methods to estimate the risk associated with each LIPI category. The analysis found that patients with an intermediate LIPI score had a 73% higher risk of death compared to those with a good score, while patients with a poor LIPI score faced nearly four times the risk of death. Similarly, poor LIPI was associated with approximately 2.7 times higher risk of disease progression, with consistent results across different cancer types and treatment approaches including immunotherapy. These findings suggest that LIPI — which can be calculated from routine blood tests — could serve as a practical and widely available tool to help oncologists identify high-risk patients, guide treatment decisions, and improve the design of clinical trials.
Frontiers in immunology
Source →The role of AI-powered molecular profiling in the diagnosis and management of cancers of unknown primary: a case report and literature review.
Esmail A, et al
Researchers from Frontiers in Oncology present a case report examining the use of artificial intelligence-powered molecular profiling to diagnose a notoriously difficult cancer type known as cancer of unknown primary, where standard tests fail to identify where the tumor originated. A 69-year-old man with a prior history of prostate cancer was initially misdiagnosed and received ineffective treatments after conventional imaging and tissue analysis could not pinpoint his cancer's origin. An AI-driven tool called Molecular Intelligence assigned an 82% probability that the cancer was renal cell carcinoma, leading physicians to perform a targeted biopsy that confirmed this diagnosis along with an aggressive sarcomatoid variant. Armed with a confirmed diagnosis, the clinical team initiated a modern immunotherapy-and-targeted-therapy combination of pembrolizumab and axitinib, which proved effective where earlier regimens had failed. This case illustrates how AI-powered molecular profiling can break diagnostic deadlocks, enabling truly personalized cancer treatment and potentially saving patients from months of ineffective therapy.
Frontiers in oncology
Source →Indole-3-propionic acid suppresses prostate cancer by inducing cell cycle arrest and apoptosis associated with p53 activation.
Huang Y, et al
Researchers investigated the antitumor potential of indole-3-propionic acid (IPA), a metabolite produced by gut bacteria, in prostate cancer, a disease whose advanced, castration-resistant form remains difficult to treat. Using blood samples from prostate cancer patients, the team found that IPA levels were significantly lower in patients with prostate cancer, especially those with high-grade disease, compared to men with benign prostatic hyperplasia. Laboratory experiments showed that IPA reduced cancer cell growth, migration, and invasion, while animal studies using xenograft mouse models confirmed that IPA substantially slowed tumor growth in living organisms. Molecular analysis revealed that IPA works by activating the p53-p21-RB signaling pathway, triggering cell cycle arrest and programmed cell death (apoptosis) in cancer cells. These findings suggest that IPA could serve both as a prognostic biomarker for prostate cancer severity and as a novel therapeutic agent rooted in microbiome-based medicine, opening new avenues for treatment strategies that harness the gut-tumor axis.
Frontiers in oncology
Source →ViTCNN: a robust hybrid CNN-Vision Transformer based deep learning framework for multi-disease diagnosis in women's healthcare.
Juneja S, et al
Researchers developed ViTCNN, a hybrid deep learning framework that combines the EfficientNetB0 convolutional neural network with a Vision Transformer to simultaneously detect three conditions critical to women's health: breast cancer, cervical cancer, and polycystic ovary syndrome (PCOS). The model uses a shared backbone with a multi-head output architecture, allowing it to capture both fine-grained local image features and broader contextual patterns within a single unified system. After an initial training phase of 70 epochs followed by 30 epochs of fine-tuning on thousands of annotated diagnostic images, the framework achieved accuracies of 98.82% for breast cancer, 95.96% for cervical cancer, and 98.96% for PCOS detection. These results surpass many single-disease models reported in the literature, demonstrating the viability of a unified multi-disease diagnostic tool. For clinicians and healthcare systems, a single robust model capable of screening multiple conditions could reduce diagnostic workload, lower costs, and improve early detection rates for women worldwide. The authors plan to expand the dataset and conduct clinical validation to prepare the framework for real-world deployment.
Frontiers in oncology
Source →[Risk assessment of subsequent drug resistance in patients with chemoresistant gestational trophoblastic neoplasia].
Mao MY, et al
Researchers at Peking Union Medical College Hospital conducted a retrospective study to identify risk factors for subsequent drug resistance in patients with chemoresistant gestational trophoblastic neoplasia (GTN), a rare but treatable pregnancy-related cancer. The study compared 248 chemoresistant GTN patients with 438 primary-treatment patients treated between 2002 and 2019, finding that chemoresistant patients had significantly higher FIGO prognostic scores, resistance rates (34.7% vs 14.8%), and disease progression rates than those receiving first-line treatment. A FIGO prognostic score of 10 or above emerged as a key threshold, with patients scoring at or above this cutoff facing substantially worse outcomes including higher rates of re-resistance and disease progression. Using univariate and multivariate logistic regression, the investigators identified independent risk factors for subsequent resistance and developed a novel scoring system to help clinicians predict which chemoresistant patients are at highest risk of further treatment failure. These findings are clinically important because they provide oncologists with a practical risk-stratification tool to guide treatment decisions and potentially improve outcomes in this challenging patient population.
Zhonghua fu chan ke za zhi
Source →