Adversarial regularized autoencoder graph neural network for microbe-disease associations prediction DOI Creative Commons

Limuxuan He,

Quan Zou, Qi Dai

и другие.

Briefings in Bioinformatics, Год журнала: 2024, Номер 25(6)

Опубликована: Сен. 23, 2024

Abstract Background Microorganisms inhabit various regions of the human body and significantly contribute to numerous diseases. Predicting associations between microbes diseases is crucial for understanding pathogenic mechanisms informing prevention treatment strategies. Biological experiments determine these are time-consuming costly. Therefore, integrating deep learning with biological networks can efficiently identify potential microbe-disease on a large scale. Methods We propose an adversarial regularized autoencoder graph neural network algorithm, named Stacked Adversarial Regularization Microbe-Disease Associations Prediction (SARMDA), predicting First, we integrate topological structural similarity functional metrics construct heterogeneous network. Then, utilizing based GraphSAGE, learn both attribute representations nodes within constructed Finally, introduce embedding model address inherent limitations traditional GraphSAGE autoencoders in capturing global information. Results Under five-fold cross-validation pairs, SARMDA was compared eight advanced methods using Human Association Database (HMDAD) Disbiome databases. The best area under ROC curve (AUC) achieved by HMDAD 0.9891$\pm$0.0057, precision-recall (AUPR) 0.9902$\pm$0.0128. On dataset, AUC 0.9328$\pm$0.0072, AUPR 0.9233$\pm$0.0089, outperforming other MDAs prediction methods. Furthermore, effectiveness our demonstrated through detailed analysis asthma inflammatory bowel disease cases.

Язык: Английский

Overcoming immunotherapy resistance in gastric cancer: insights into mechanisms and emerging strategies DOI Creative Commons

D.Y. Luo,

Jing Zhou, Shuiliang Ruan

и другие.

Cell Death and Disease, Год журнала: 2025, Номер 16(1)

Опубликована: Фев. 7, 2025

Abstract Gastric cancer (GC) remains a leading cause of cancer-related mortality worldwide, with limited treatment options in advanced stages. Immunotherapy, particularly immune checkpoint inhibitors (ICIs) targeting PD1/PD-L1, has emerged as promising therapeutic approach. However, significant proportion patients exhibit primary or acquired resistance, limiting the overall efficacy immunotherapy. This review provides comprehensive analysis mechanisms underlying immunotherapy resistance GC, including role tumor microenvironment, dynamic PD-L1 expression, compensatory activation other checkpoints, and genomic instability. Furthermore, explores GC-specific factors such molecular subtypes, unique evasion mechanisms, impact Helicobacter pylori infection. We also discuss emerging strategies to overcome combination therapies, novel immunotherapeutic approaches, personalized based on genomics microenvironment. By highlighting these key areas, this aims inform future research directions clinical practice, ultimately improving outcomes for GC undergoing

Язык: Английский

Процитировано

2
Role of human microbiota in facilitating the metastatic journey of cancer cells DOI

Himisa Shah,

Premal H. Patel,

Aritro Nath

и другие.

Naunyn-Schmiedeberg s Archives of Pharmacology, Год журнала: 2025, Номер unknown

Опубликована: Март 12, 2025

Язык: Английский

Процитировано

0
Proton pump inhibitors reduce nivolumab efficacy in unresectable advanced or recurrent gastric cancer DOI

Masahito Shibano,

Masaya Takahashi,

Hitomi Nakatsukasa

и другие.

Immunotherapy, Год журнала: 2025, Номер unknown, С. 1 - 8

Опубликована: Апрель 14, 2025

Proton pump inhibitors (PPI) have been shown to decrease the efficacy of immune checkpoint in patients with various cancer types. However, there are few reports on their effect gastric (GC). Therefore, we investigated nivolumab GC receiving PPI. This retrospective study analyzed data who received monotherapy for unresectable advanced or recurrent at Osaka Metropolitan University Hospital between September 2017 and December 2021. The primary secondary endpoints were progression-free survival (PFS) overall (OS), respectively. PPI use was defined as within 30 days before after initiation monotherapy. Seventy-seven eligible included this analysis. PPIs used 33 patients, while 36 had a previous gastrectomy. Multivariate analysis revealed that only an independent predictor PFS (hazard ratio [HR] 1.93, 95% confidence interval [CI] 1.03-3.64, p = 0.042). Contrastingly, not OS. may reduce nivolumab, should be carefully considered nivolumab.

Язык: Английский

Процитировано

0
Impact ofHelicobacter pylorion immunotherapy in gastric cancer DOI Creative Commons
Jing Li, Zongyin Wu,

Rong Lin

и другие.

Journal for ImmunoTherapy of Cancer, Год журнала: 2024, Номер 12(10), С. e010354 - e010354

Опубликована: Окт. 1, 2024

This study reviews the contrasting finding regarding impact of

Язык: Английский

Процитировано

2
Adversarial regularized autoencoder graph neural network for microbe-disease associations prediction DOI Creative Commons

Limuxuan He,

Quan Zou, Qi Dai

и другие.

Briefings in Bioinformatics, Год журнала: 2024, Номер 25(6)

Опубликована: Сен. 23, 2024

Abstract Background Microorganisms inhabit various regions of the human body and significantly contribute to numerous diseases. Predicting associations between microbes diseases is crucial for understanding pathogenic mechanisms informing prevention treatment strategies. Biological experiments determine these are time-consuming costly. Therefore, integrating deep learning with biological networks can efficiently identify potential microbe-disease on a large scale. Methods We propose an adversarial regularized autoencoder graph neural network algorithm, named Stacked Adversarial Regularization Microbe-Disease Associations Prediction (SARMDA), predicting First, we integrate topological structural similarity functional metrics construct heterogeneous network. Then, utilizing based GraphSAGE, learn both attribute representations nodes within constructed Finally, introduce embedding model address inherent limitations traditional GraphSAGE autoencoders in capturing global information. Results Under five-fold cross-validation pairs, SARMDA was compared eight advanced methods using Human Association Database (HMDAD) Disbiome databases. The best area under ROC curve (AUC) achieved by HMDAD 0.9891$\pm$0.0057, precision-recall (AUPR) 0.9902$\pm$0.0128. On dataset, AUC 0.9328$\pm$0.0072, AUPR 0.9233$\pm$0.0089, outperforming other MDAs prediction methods. Furthermore, effectiveness our demonstrated through detailed analysis asthma inflammatory bowel disease cases.

Язык: Английский

Процитировано

1