Hasil untuk "Neoplasms. Tumors. Oncology. Including cancer and carcinogens"

Dapeng Li, Jing Lu, Ran Zhu et al.

Bladder cancer remains a leading cause of cancer-related mortality with limited therapeutic options. This study investigates the antitumor efficacy and mechanism of Ad-VT, a dual-specific oncolytic adenovirus expressing apoptin under the hTERT promoter, in bladder cancer. In vitro, Ad-VT selectively killed bladder cancer cells (UM-UC-3, T24, 5637, RT4) while sparing normal urothelial cells (SV-HUC-1), showing dose-dependent cytotoxicity (70 % inhibition at 100 MOI in 5637 cells). It induced G2/M phase arrest via downregulation of cyclin B1/cdc2 and upregulation of p-cdc2/p21. Mechanistically, Ad-VT elevated cAMP levels, activating the AMPK-Raptor-mTOR pathway. This was confirmed by pathway inhibitors (Dorsomorphin, ESI-09) and siRNA knockdown, which reversed cell cycle arrest and reduced cytotoxicity. In vivo, intratumoral Ad-VT injection suppressed UM-UC-3 xenograft growth, enhanced survival, and increased apoptosis while reducing proliferation. Crucially, AMPK inhibition attenuated Ad-VT's antitumor effects. These results demonstrate that Ad-VT exerts potent, tumor-selective activity against bladder cancer by inducing cAMP-dependent AMPK-Raptor-mTOR signaling and G2/M arrest, supporting its therapeutic potential.

Suong Duong, Danny Lee, Carri Glide-Hurst et al.

BackgroundMulti-institutional clinical trials frequently use MRI imaging for critical decisions and guidance for medical treatments. Collecting and analyzing images produced by various MR vendors and models is quite difficult since image quality can be highly variable. No unifying quality control targeting protocol studies exists to ensure MRI images used in that study are comparable. This project will investigate variations between imaging sessions and between various scanners using radiomic parameters from prototype MRI QA phantom.PurposeTo develop a 3D radiomic phantom for quantifying radiomic feature consistency between MRI scanners across multi-institutions.MethodsThe prototype phantom consists of five 3D-printed objects (3 grid and 2 egg-shape) using Polylactic Acid (PLA) with/without 20% wood particles placed in a water container. The grid objects consisted of PLA scaffolding with 245 cubic voids (flood-filled by water) stacked in 7rows x 7columns x 5layers with volumes of 3x3x3 mm3 or 5x5x5 mm3, and scaffolding thickness of 1mm or 2mm. The egg-shaped objects are 5cm long with a 2cm or 4cm maximum diameter, filled with vitamin D3-capsules and olive-oil. It was scanned 10 times using T1- and T2-weighted sequences on Philips (1.5T Elekta Unity), GE (1.5T, Signa Artist), Siemens (1.5T MAGNETOM Sola), and Philips (1.5T Ingenia) across four institutions. TrueFISP and T2w sequences were used on ViewRay (0.35T MRIdian) scanners at two institutions. Per object, 107 radiomic features were extracted using the Pyradiomics extension in 3D Slicer. Coefficients of Variation (CV) of individual radiomic features were compared across 10 scans acquired on each scanner and used to compare radiomic feature consistency between objects and MRI scanners.ResultsThe radiomic feature consistency varied across objects with less reproducibility for the egg-shaped objects and more reproducibility for the grid objects, with slightly better reproducibility for T1w than T2w sequences. The GE scanner demonstrated better reproducibility than the other scanners. Both ViewRay scanners showed consistency for acquisitions with the TrueFISP sequence; the median CV of 107 radiomic features between objects was <10%). The consistency was summarized in a heat map.ConclusionSome radiomic features showed significant intra-scanner variations. This study demonstrated that a standardized radiomic phantom is required to characterize individual scanners and MR sequences for establishing the baseline of radiomic features, which could be important for multi-institutional radiomic studies using MRI.

Yibin Zheng, Hang Cai, Hongbin Huang et al.

Abstract Disulfidptosis was reported to be associated with the malignant progression of various tumors. This study was aimed to investigate the prognostic significance of disulfidptosis-related genes (DRGs) in osteosarcoma (OS). Ten previously reported core disulfidptosis genes were used for consensus clustering and WGCNA analyses. A total of 338 disulfidptosis-related genes (DRGs) were identified. Then, uni-COX, LASSO and multi-COX analyses were conducted, identifying 5 prognosis-related DRGs (BTN3A1, CEBPA, KCNAB2, TBX21, and MYC). A prognostic DRGs risk signature based on the five genes were constructed and validated. OS patients were divided into high and low risk groups by risk scores. K-M plots and t-ROC curves showed that patients with high-risk scores had worse prognosis. Patients in the high-risk group had lower abundance of immune checkpoint-related genes, including CD274 (PD-L1), LAG3, PDCD1LG2 (PD-L2), and BTLA. Besides, patients in the high-risk group exhibited lower IC50 values for vorinostat, elesclomol, OSI-906, pyrimethamine, thapsigargin, and doxorubicin, but a higher IC50 value for cisplatin, compared to those in the low-risk group, indicating differential drug sensitivities. Additionally, analysis of immune checkpoint blockade (ICB) response revealed that patients in the high-risk group had a lower predicted response rate to immunotherapy. The mRNA expression levels of 4 DRGs including BTN3A1, KCNAB2, TBX21 and CEBPA in OS cells were significantly lower than those in hFOB1.19 cells. Subsequent experiments revealed that BTN3A1 protein was expressed at low levels in OS cells. Furthermore, overexpression of BTN3A1 significantly suppressed OS cell proliferation, migration, and invasion. In summary, we established a robust DRGs signature comprising BTN3A1, CEBPA, KCNAB2, TBX21, and MYC, which showed strong prognostic value and predictive potential for immune status and drug sensitivity in OS. Notably, functional experiments confirmed that BTN3A1 acted as a tumor suppressor in OS, highlighting it as a promising therapeutic target.

Carlijn Lems, Leslie Tessier, John-Melle Bokhorst et al.

Automated semantic segmentation of whole-slide images (WSIs) stained with hematoxylin and eosin (H&E) is essential for large-scale artificial intelligence-based biomarker analysis in breast cancer. However, existing public datasets for breast cancer segmentation lack the morphological diversity needed to support model generalizability and robust biomarker validation across heterogeneous patient cohorts. We introduce BrEast cancEr hisTopathoLogy sEgmentation (BEETLE), a dataset for multiclass semantic segmentation of H&E-stained breast cancer WSIs. It consists of 587 biopsies and resections from three collaborating clinical centers and two public datasets, digitized using seven scanners, and covers all molecular subtypes and histological grades. Using diverse annotation strategies, we collected annotations across four classes - invasive epithelium, non-invasive epithelium, necrosis, and other - with particular focus on morphologies underrepresented in existing datasets, such as ductal carcinoma in situ and dispersed lobular tumor cells. The dataset's diversity and relevance to the rapidly growing field of automated biomarker quantification in breast cancer ensure its high potential for reuse. Finally, we provide a well-curated, multicentric external evaluation set to enable standardized benchmarking of breast cancer segmentation models.

Vo Duy Long, Dang Quang Thong, Tran Quang Dat et al.

Abstract Background No studies on neoadjuvant chemotherapy for gastric cancer (GC) with T4b stage were reported. This study aimed to assess the effectiveness of neoadjuvant chemotherapy using DCS regimen (docetaxel, cisplatin, and S-1) for GC with T4b stage. Methods Forty-three patients diagnosed GC with surgical or clinical T4b stage received three or four preoperative cycles of DCS therapy followed by gastrectomy and lymphadenectomy between Jan-2018 and Dec-2022. Short-tern outcomes including tumor response, completion of neoadjuvant chemotherapy, toxicity and adverse events, rate of treatment-related death, R0 resection, rate of complete adjuvant chemotherapy and short-term surgical results were investigated. The oncologic outcomes comprised 3-year OS and 3-year disease-free survival (DFS). Results A total of 43 patients with T4b gastric cancer were included in the analysis. Among them, twenty-five patients underwent gastrectomy and lymphadenectomy. The completion rate of neoadjuvant chemotherapy was 88.4%, including 4 cycles of 51.2% and 3 cycles of 37.2%. The disease-control and clinical response rate were 88.4% and 58.1%, respectively. During preoperative chemotherapy, grade 3/4 neutropenia occurred in 20.9%, anemia in 13.9%, hyponatremia in 4.8%, and vomiting in 2.3%. Pathologic complete response was achieved in 8.0%. After surgery, no patient experienced severe complications (Clavien Dindo > = 3). The R0 resection rate was 72.0% and the rate of complete adjuvant chemotherapy was 83.3%. The 3-year OS and DFS rates were 49% and 38%, respectively. Conclusions Neoadjuvant chemotherapy with DCS regimen demonstrated a high tolerance, high tumor response rate, high complete adjuvant chemotherapy rate and satisfactory 3-year survival outcomes. Three- or four-course of preoperative DCS regimen is a promising approach for GC with T4b stage.

Anurag Kumar Singh, Kaivalya Walavalkar, Daniele Tavernari et al.

Human papillomavirus (HPV) infections are the primary drivers of cervical cancers, and often HPV DNA gets integrated into the host genome. Although the oncogenic impact of HPV encoded genes is relatively well known, the cis‐regulatory effect of integrated HPV DNA on host chromatin structure and gene regulation remains less understood. We investigated genome‐wide patterns of HPV integrations and associated host gene expression changes in the context of host chromatin states and topologically associating domains (TADs). HPV integrations were significantly enriched in active chromatin regions and depleted in inactive ones. Interestingly, regardless of chromatin state, genomic regions flanking HPV integrations showed transcriptional upregulation. Nevertheless, upregulation (both local and long‐range) was mostly confined to TADs with integration, but not affecting adjacent TADs. Few TADs showed recurrent integrations associated with overexpression of oncogenes within them (e.g. MYC, PVT1, TP63 and ERBB2) regardless of proximity. Hi‐C and 4C‐seq analyses in cervical cancer cell line (HeLa) demonstrated chromatin looping interactions between integrated HPV and MYC/PVT1 regions (~ 500 kb apart), leading to allele‐specific overexpression. Based on these, we propose HPV integrations can trigger multimodal oncogenic activation to promote cancer progression.

Davide Belluomo, Tiziana Calamoneri, Giacomo Paesani et al.

We present a new unified graph-based representation of medical data, combining genetic information and medical records of patients with medical knowledge via a unique knowledge graph. This approach allows us to infer meaningful information and explanations that would be unavailable by looking at each data set separately. The systematic use of different databases, managed throughout the built knowledge graph, gives new insights toward a better understanding of oncology medicine. Indeed, we reduce some useful medical tasks to well-known problems in theoretical computer science for which efficient algorithms exist.

Daniele Perlo, Georgia Kanli, Selma Boudissa et al.

This research paper presents a novel approach to the prediction of hypoxia in brain tumors, using multi-parametric Magnetic Resonance Imaging (MRI). Hypoxia, a condition characterized by low oxygen levels, is a common feature of malignant brain tumors associated with poor prognosis. Fluoromisonidazole Positron Emission Tomography (FMISO PET) is a well-established method for detecting hypoxia in vivo, but it is expensive and not widely available. Our study proposes the use of MRI, a more accessible and cost-effective imaging modality, to predict FMISO PET signals. We investigate deep learning models (DL) trained on the ACRIN 6684 dataset, a resource that contains paired MRI and FMISO PET images from patients with brain tumors. Our trained models effectively learn the complex relationships between the MRI features and the corresponding FMISO PET signals, thereby enabling the prediction of hypoxia from MRI scans alone. The results show a strong correlation between the predicted and actual FMISO PET signals, with an overall PSNR score above 29.6 and a SSIM score greater than 0.94, confirming MRI as a promising option for hypoxia prediction in brain tumors. This approach could significantly improve the accessibility of hypoxia detection in clinical settings, with the potential for more timely and targeted treatments.

Marina Chugunova, Hangjie Ji, Roman Taranets et al.

In this work, we focus on the analytical and numerical study of a mathematical model for brain tumors with radiotherapy influence. Under certain assumptions on the given data in the model, we prove existence and uniqueness of a weak nonnegative (biological relevant) solution. Then, assuming only more regular initial data, we obtain the extra regularity of this solution. Besides, we analyze the optimal control of the advection coefficient responding for the radiotherapy effect on the tumor cell population. Finally, we provide numerical illustration to all obtained analytical results.

Milad Yousefi, Shadi Farabi Maleki, Ali Jafarizadeh et al.

Thyroid cancer is an increasing global health concern that requires advanced diagnostic methods. The application of AI and radiomics to thyroid cancer diagnosis is examined in this review. A review of multiple databases was conducted in compliance with PRISMA guidelines until October 2023. A combination of keywords led to the discovery of an English academic publication on thyroid cancer and related subjects. 267 papers were returned from the original search after 109 duplicates were removed. Relevant studies were selected according to predetermined criteria after 124 articles were eliminated based on an examination of their abstract and title. After the comprehensive analysis, an additional six studies were excluded. Among the 28 included studies, radiomics analysis, which incorporates ultrasound (US) images, demonstrated its effectiveness in diagnosing thyroid cancer. Various results were noted, some of the studies presenting new strategies that outperformed the status quo. The literature has emphasized various challenges faced by AI models, including interpretability issues, dataset constraints, and operator dependence. The synthesized findings of the 28 included studies mentioned the need for standardization efforts and prospective multicenter studies to address these concerns. Furthermore, approaches to overcome these obstacles were identified, such as advances in explainable AI technology and personalized medicine techniques. The review focuses on how AI and radiomics could transform the diagnosis and treatment of thyroid cancer. Despite challenges, future research on multidisciplinary cooperation, clinical applicability validation, and algorithm improvement holds the potential to improve patient outcomes and diagnostic precision in the treatment of thyroid cancer.

Masoud Tafavvoghi, Anders Sildnes, Mehrdad Rakaee et al.

Classifying breast cancer molecular subtypes is crucial for tailoring treatment strategies. While immunohistochemistry (IHC) and gene expression profiling are standard methods for molecular subtyping, IHC can be subjective, and gene profiling is costly and not widely accessible in many regions. Previous approaches have highlighted the potential application of deep learning models on H&E-stained whole slide images (WSI) for molecular subtyping, but these efforts vary in their methods, datasets, and reported performance. In this work, we investigated whether H&E-stained WSIs could be solely leveraged to predict breast cancer molecular subtypes (luminal A, B, HER2-enriched, and Basal). We used 1,433 WSIs of breast cancer in a two-step pipeline: first, classifying tumor and non-tumor tiles to use only the tumor regions for molecular subtyping; and second, employing a One-vs-Rest (OvR) strategy to train four binary OvR classifiers and aggregating their results using an eXtreme Gradient Boosting (XGBoost) model. The pipeline was tested on 221 hold-out WSIs, achieving an overall macro F1 score of 0.95 for tumor detection and 0.73 for molecular subtyping. Our findings suggest that, with further validation, supervised deep learning models could serve as supportive tools for molecular subtyping in breast cancer. Our codes are made available to facilitate ongoing research and development.

Alice Oh, Inyoung Noh, Jian Choo et al.

Brain tumor detection and classification are critical tasks in medical image analysis, particularly in early-stage diagnosis, where accurate and timely detection can significantly improve treatment outcomes. In this study, we apply various statistical and machine learning models to detect and classify brain tumors using brain MRI images. We explore a variety of statistical models including linear, logistic, and Bayesian regressions, and the machine learning models including decision tree, random forest, single-layer perceptron, multi-layer perceptron, convolutional neural network (CNN), recurrent neural network, and long short-term memory. Our findings show that CNN outperforms other models, achieving the best performance. Additionally, we confirm that the CNN model can also work for multi-class classification, distinguishing between four categories of brain MRI images such as normal, glioma, meningioma, and pituitary tumor images. This study demonstrates that machine learning approaches are suitable for brain tumor detection and classification, facilitating real-world medical applications in assisting radiologists with early and accurate diagnosis.

Mohammad Amin Abazari, M. Soltani, Faezeh Eydi et al.

$^{18}$F-Fluoromisonidazole ($^{18}$F-FMISO) is a highly promising positron emission tomography radiopharmaceutical for identifying hypoxic regions in solid tumors. This research employs spatiotemporal multi-scale mathematical modeling to explore how different levels of angiogenesis influence the transport of radiopharmaceuticals within tumors. In this study, two tumor geometries with heterogeneous and uniform distributions of capillary networks were employed to incorporate varying degrees of microvascular density. The synthetic image of the heterogeneous and vascularized tumor was generated by simulating the angiogenesis process. The proposed multi-scale spatiotemporal model accounts for intricate physiological and biochemical factors within the tumor microenvironment, such as the transvascular transport of the radiopharmaceutical agent, its movement into the interstitial space by diffusion and convection mechanisms, and ultimately its uptake by tumor cells. Results showed that both quantitative and semi-quantitative metrics of $^{18}$F-FMISO uptake differ spatially and temporally at different stages during tumor growth. The presence of a high microvascular density in uniformly vascularized tumor increases cellular uptake, as it allows for more efficient release and rapid distribution of radiopharmaceutical molecules. This results in enhanced uptake compared to the heterogeneous vascularized tumor. In both heterogeneous and uniform distribution of microvessels in tumors, the diffusion transport mechanism has a more pronounced than convection. The findings of this study shed light on the transport phenomena behind $^{18}$F-FMISO radiopharmaceutical distribution and its delivery in the tumor microenvironment, aiding oncologists in their routine decision-making processes.

Reza Babaei, Samuel Cheng, Theresa Thai et al.

Despite the advances in medicine, cancer has remained a formidable challenge. Particularly in the case of pancreatic tumors, characterized by their diversity and late diagnosis, early detection poses a significant challenge crucial for effective treatment. The advancement of deep learning techniques, particularly supervised algorithms, has significantly propelled pancreatic tumor detection in the medical field. However, supervised deep learning approaches necessitate extensive labeled medical images for training, yet acquiring such annotations is both limited and costly. Conversely, weakly supervised anomaly detection methods, requiring only image-level annotations, have garnered interest. Existing methodologies predominantly hinge on generative adversarial networks (GANs) or autoencoder models, which can pose complexity in training and, these models may face difficulties in accurately preserving fine image details. This research presents a novel approach to pancreatic tumor detection, employing weak supervision anomaly detection through denoising diffusion algorithms. By incorporating a deterministic iterative process of adding and removing noise along with classifier guidance, the method enables seamless translation of images between diseased and healthy subjects, resulting in detailed anomaly maps without requiring complex training protocols and segmentation masks. This study explores denoising diffusion models as a recent advancement over traditional generative models like GANs, contributing to the field of pancreatic tumor detection. Recognizing the low survival rates of pancreatic cancer, this study emphasizes the need for continued research to leverage diffusion models' efficiency in medical segmentation tasks.

Aldo Caltavituro, Roberto Buonaiuto, Fabio Salomone et al.

BackgroundEwing sarcoma (ES) represents the second most common malignant bone tumor in children and young adults. ES is not a frequent finding in sites different from the skeletal. Common sites of appearance of ES are lower extremities, the pelvis, paravertebral spaces and head and neck. Primary extraskeletal ES located in the anterior mediastinum are very rare. These neoplasms should be discussed in specialized contests with a high volume of patients treated. Here, we present an uncommon mediastinal mass challenging in its characterization and management.Case descriptionA thirty-year-old woman performed a thoracic CT scan for dyspnea and persistent cough. Imaging showed a solid mass of 14 x 11 cm involving the left thorax with mediastinal deviation to the right side. Patient underwent an en bloc resection of the mass. Initial histological examination was suggestive for B3 thymoma/thymic carcinoma. Patient was then referred to our rare tumor reference center where a histological review excluded the diagnosis of thymic/thymoma neoplasms meanwhile a third revision assessed a diagnosis of ES. Patient refused adjuvant chemotherapy due to her desire of maternity and radiation therapy was not indicated because surgery was performed too many months earlier. A close follow-up was considered. After a few months the patient relapsed and first line chemotherapy was proposed. She reached a complete response at the first evaluation maintained also at the end of the protocol. In order to consolidate the obtained response, high dose chemotherapy followed by autologous stem cell transplantation (HDCT/ASCT) was suggested and the patient agreed.ConclusionsThis case underlined that, potentially, ES can arise from any soft tissue site in the body, even in rare sites such as mediastinum. The evaluation of expert centers was critical to establish a correct diagnosis and therapeutic approach in this complex case. Taking into account the time lasting from the diagnosis and the aggressiveness of this kind of neoplasm, frequently relapsing, the patient after a multidisciplinary discussion was a candidate for a multimodal treatment.

Lindor Qunaj, Michael S. May, Alfred I. Neugut et al.

KRAS G12C mutations are critical in the pathogenesis of multiple cancer types, including non-small cell lung (NSCLC), pancreatic ductal adenocarcinoma (PDAC), and colorectal (CRC) cancers. As such, they have increasingly become a target of novel therapies in the management of these malignancies. However, the therapeutic success of KRAS G12C inhibitors to date has been far more limited in CRC and PDAC than NSCLC. In this review, we briefly summarize the biochemistry of KRAS targeting and treatment resistance, highlight differences in the epidemiology of various G12C-mutated cancers, and provide an overview of the published data on KRAS G12C inhibitors for various indications. We conclude with a summary of ongoing clinical trials in G12C-mutant CRC and a discussion of future directions in the management of this disease. KRAS G12C mutation, targeted therapies, colorectal cancer, non-small cell lung cancer, pancreatic cancer, drug development.

Mark Laurie, James Lu

While tumor dynamic modeling has been widely applied to support the development of oncology drugs, there remains a need to increase predictivity, enable personalized therapy, and improve decision-making. We propose the use of Tumor Dynamic Neural-ODE (TDNODE) as a pharmacology-informed neural network to enable model discovery from longitudinal tumor size data. We show that TDNODE overcomes a key limitation of existing models in its ability to make unbiased predictions from truncated data. The encoder-decoder architecture is designed to express an underlying dynamical law which possesses the fundamental property of generalized homogeneity with respect to time. Thus, the modeling formalism enables the encoder output to be interpreted as kinetic rate metrics, with inverse time as the physical unit. We show that the generated metrics can be used to predict patients' overall survival (OS) with high accuracy. The proposed modeling formalism provides a principled way to integrate multimodal dynamical datasets in oncology disease modeling.

Isabel González-Crespo, Faustino Gómez, Óscar López Pouso et al.

FLASH radiotherapy (FLASH-RT) has shown the potential to spare normal tissue while seemingly maintaining the effectiveness of conventional radiotherapy (CONV-RT). It has been suggested that the protective effect arises from the radiolytic oxygen depletion (ROD) caused by FLASH-RT, but it is not entirely clear why this protective effect is not observed in tumors. Iso-effectiveness has been experimentally observed in time-volume curves of preclinical tumors irradiated with FLASH and conventional radiotherapy, but it may not translate to clinical trials, where tumor control probability (TCP) is typically the investigated endpoint. In this work, we used mathematical models to investigate the iso-effectiveness of FLASH-RT/CONV-RT on tumors, focusing on the role of ROD. We used a spatiotemporal reaction-diffusion model, including ROD, to simulate tumor oxygenation. From those oxygen distributions we obtained surviving fractions (SFs), using the linear-quadratic model with oxygen enhancement ratios (OER). We then used the calculated SFs to describe the evolution of preclinical tumor volumes through a mathematical model of tumor response. We also calculated TCPs using the Poisson-LQ approach. Our study suggests that ROD causes differences in SF between FLASH-RT and CONV-RT, especially in low $α$/$β$ and poorly oxygenated cells. These changes do not lead to significant differences in the evolution of preclinical tumors. However, when extrapolating this effect to TCP curves, we observed important differences between both techniques (TCP is lower in FLASH-RT). Nonetheless, it cannot be discarded that other effects not modeled in this work could contribute to tumor control and maintain the iso-effectiveness of FLASH-RT.

Zihao Chen, Jia Lu, Xing-Ming Zhao et al.

Cancer is a systemic heterogeneous disease involving complex molecular networks. Tumor formation involves epithelial-mesenchymal transition (EMT), which promotes both metastasis and plasticity of cancer cells. Recent experiments proposed that cancer cells can be transformed into adipocytes with combination drugs. However, the underlying mechanisms for how these drugs work from molecular network perspective remain elusive. To reveal the mechanism of cancer-adipose conversion (CAC), we adopt a systems biology approach by combing mathematical modeling and molecular experiments based on the underlying molecular regulatory network. We identified four types of attractors which correspond to epithelial (E), mesenchymal (M), adipose (A) and partial/intermediate EMT (P) cell states on the CAC landscape. Landscape and transition path results illustrate that the intermediate states play critical roles in cancer to adipose transition. Through a landscape control strategy, we identified two new therapeutic strategies for drug combinations to promote CAC. We further verified these predictions by molecular experiments in different cell lines. Our combined computational and experimental approach provides a powerful tool to explore molecular mechanisms for cell fate transitions in cancer networks. Our results revealed the underlying mechanism for intermediate cell states governing the CAC, and identified new potential drug combinations to induce cancer adipogenesis.

Michail Chatzianastasis, Michalis Vazirgiannis, Zijun Zhang

The identification of cancer genes is a critical yet challenging problem in cancer genomics research. Existing computational methods, including deep graph neural networks, fail to exploit the multilayered gene-gene interactions or provide limited explanation for their predictions. These methods are restricted to a single biological network, which cannot capture the full complexity of tumorigenesis. Models trained on different biological networks often yield different and even opposite cancer gene predictions, hindering their trustworthy adaptation. Here, we introduce an Explainable Multilayer Graph Neural Network (EMGNN) approach to identify cancer genes by leveraging multiple genegene interaction networks and pan-cancer multi-omics data. Unlike conventional graph learning on a single biological network, EMGNN uses a multilayered graph neural network to learn from multiple biological networks for accurate cancer gene prediction. Our method consistently outperforms all existing methods, with an average 7.15% improvement in area under the precision-recall curve (AUPR) over the current state-of-the-art method. Importantly, EMGNN integrated multiple graphs to prioritize newly predicted cancer genes with conflicting predictions from single biological networks. For each prediction, EMGNN provided valuable biological insights via both model-level feature importance explanations and molecular-level gene set enrichment analysis. Overall, EMGNN offers a powerful new paradigm of graph learning through modeling the multilayered topological gene relationships and provides a valuable tool for cancer genomics research.