Biological informed graph neural network for tumor mutation burden prediction and immunotherapy-related pathway analysis in gastric cancer
© 2023 The Authors..
Tumor mutation burden (TMB) has emerged as an essential biomarker for assessing the efficacy of cancer immunotherapy. However, due to the inherent complexity of tumors, TMB is not always correlated with the responsiveness of immune checkpoint inhibitors (ICIs). Thus, refining the interpretation and contextualization of TMB is a requisite for enhancing clinical outcomes. In this study, we conducted a comprehensive investigation of the relationship between TMB and multi-omics data across 33 human cancer types. Our analysis revealed distinct biological changes associated with varying TMB statuses in STAD, COAD, and UCEC. While multi-omics data offer an opportunity to dissect the intricacies of tumors, extracting meaningful biological insights from such massive information remains a formidable challenge. To address this, we developed and implemented the PGLCN, a biologically informed graph neural network based on pathway interaction information. This model facilitates the stratification of patients into subgroups with distinct TMB statuses and enables the evaluation of driver biological processes through enhanced interpretability. By integrating multi-omics data for TMB prediction, our PGLCN model outperformed previous traditional machine learning methodologies, demonstrating superior TMB status prediction accuracy (STAD AUC: 0.976 ± 0.007; COAD AUC: 0.994 ± 0.007; UCEC AUC: 0.947 ± 0.023) and enhanced interpretability (BA-House: 1.0; BA-Community: 0.999; BA-Grid: 0.994; Tree-Cycles: 0.917; Tree-Grids: 0.867). Furthermore, the biological interpretability inherent to PGLCN identified the Toll-like receptor family and DNA repair pathways as potential combined biomarkers in conjunction with TMB status in gastric cancer. This finding suggests a potential synergistic targeting strategy with immunotherapy for gastric cancer, thus advancing the field of precision oncology.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2023 |
|---|---|
| Erschienen: |
2023 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:21 |
|---|---|
| Enthalten in: |
Computational and structural biotechnology journal - 21(2023) vom: 01., Seite 4540-4551 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Liu, Chuwei [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
Gastric cancer |
|---|
| Anmerkungen: |
Date Revised 30.10.2023 published: Electronic-eCollection Citation Status PubMed-not-MEDLINE |
|---|
| doi: |
10.1016/j.csbj.2023.09.021 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM363034625 |
|---|
| LEADER | 01000naa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM363034625 | ||
| 003 | DE-627 | ||
| 005 | 20231226092508.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 231226s2023 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1016/j.csbj.2023.09.021 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1210.xml |
| 035 | |a (DE-627)NLM363034625 | ||
| 035 | |a (NLM)37810279 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Liu, Chuwei |e verfasserin |4 aut | |
| 245 | 1 | 0 | |a Biological informed graph neural network for tumor mutation burden prediction and immunotherapy-related pathway analysis in gastric cancer |
| 264 | 1 | |c 2023 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Revised 30.10.2023 | ||
| 500 | |a published: Electronic-eCollection | ||
| 500 | |a Citation Status PubMed-not-MEDLINE | ||
| 520 | |a © 2023 The Authors. | ||
| 520 | |a Tumor mutation burden (TMB) has emerged as an essential biomarker for assessing the efficacy of cancer immunotherapy. However, due to the inherent complexity of tumors, TMB is not always correlated with the responsiveness of immune checkpoint inhibitors (ICIs). Thus, refining the interpretation and contextualization of TMB is a requisite for enhancing clinical outcomes. In this study, we conducted a comprehensive investigation of the relationship between TMB and multi-omics data across 33 human cancer types. Our analysis revealed distinct biological changes associated with varying TMB statuses in STAD, COAD, and UCEC. While multi-omics data offer an opportunity to dissect the intricacies of tumors, extracting meaningful biological insights from such massive information remains a formidable challenge. To address this, we developed and implemented the PGLCN, a biologically informed graph neural network based on pathway interaction information. This model facilitates the stratification of patients into subgroups with distinct TMB statuses and enables the evaluation of driver biological processes through enhanced interpretability. By integrating multi-omics data for TMB prediction, our PGLCN model outperformed previous traditional machine learning methodologies, demonstrating superior TMB status prediction accuracy (STAD AUC: 0.976 ± 0.007; COAD AUC: 0.994 ± 0.007; UCEC AUC: 0.947 ± 0.023) and enhanced interpretability (BA-House: 1.0; BA-Community: 0.999; BA-Grid: 0.994; Tree-Cycles: 0.917; Tree-Grids: 0.867). Furthermore, the biological interpretability inherent to PGLCN identified the Toll-like receptor family and DNA repair pathways as potential combined biomarkers in conjunction with TMB status in gastric cancer. This finding suggests a potential synergistic targeting strategy with immunotherapy for gastric cancer, thus advancing the field of precision oncology | ||
| 650 | 4 | |a Journal Article | |
| 650 | 4 | |a Gastric cancer | |
| 650 | 4 | |a Graph neural network | |
| 650 | 4 | |a Immunotherapy | |
| 650 | 4 | |a Machine learning | |
| 650 | 4 | |a Tumor mutation burden | |
| 700 | 1 | |a Wan, Arabella H |e verfasserin |4 aut | |
| 700 | 1 | |a Liang, Heng |e verfasserin |4 aut | |
| 700 | 1 | |a Sun, Lei |e verfasserin |4 aut | |
| 700 | 1 | |a Li, Jiarui |e verfasserin |4 aut | |
| 700 | 1 | |a Yang, Ranran |e verfasserin |4 aut | |
| 700 | 1 | |a Li, Qinghai |e verfasserin |4 aut | |
| 700 | 1 | |a Wu, Ruibo |e verfasserin |4 aut | |
| 700 | 1 | |a Hu, Kunhua |e verfasserin |4 aut | |
| 700 | 1 | |a Yang, Yuedong |e verfasserin |4 aut | |
| 700 | 1 | |a Cai, Shirong |e verfasserin |4 aut | |
| 700 | 1 | |a Wan, Guohui |e verfasserin |4 aut | |
| 700 | 1 | |a He, Weiling |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t Computational and structural biotechnology journal |d 2012 |g 21(2023) vom: 01., Seite 4540-4551 |w (DE-627)NLM218960549 |x 2001-0370 |7 nnns |
| 773 | 1 | 8 | |g volume:21 |g year:2023 |g day:01 |g pages:4540-4551 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1016/j.csbj.2023.09.021 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 21 |j 2023 |b 01 |h 4540-4551 | ||