Leukemia research · 2026
Integrative transcriptomic and machine learning analysis identifies core immune genes and pathways driving graft-versus-host disease after allogeneic hematopoietic stem cell transplantation.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Lu Wei, He Zhipeng, Huang Xianbao
Аннотация
BACKGROUND: Graft-versus-host disease (GVHD), a major adverse event following allogeneic hematopoietic stem cell transplantation (allo-HSCT), commonly affects leukemia patients and is associated with reduced survival and impaired quality of life. Early prediction of GVHD remains a major clinical challenge.
METHODS: This study combined transcriptomic profiles from several GEO datasets of leukemia patients after transplantation. Feature genes were selected through expression pattern analysis combined with unsupervised clustering. Machine learning algorithms were applied to construct predictive models for GVHD, with model performance evaluated by ROC curves in both training and external validation cohorts. Additionally, functional enrichment analyses, including GO, KEGG, GSEA, and GSVA, were performed to investigate potential biological mechanisms. In the final step, we explored the patterns of immune cell infiltration and their relationships with the core genes.
RESULTS: A machine learning model based on 17 core genes was developed, with Ridge regression showing the best performance (AUC = 0.955). Key genes included HLA-DQA1, IL2RA, SELPLG, GPR65, and RUVBL1. Functional enrichment revealed involvement in immune regulation, cytokine production, and hematopoietic pathways. GSEA and GSVA analyses demonstrated significant activation of pathways such as inositol phosphate metabolism, ABC transporters, and Th1/Th2 cell differentiation in GVHD patients. Analysis of immune infiltration revealed elevated levels of follicular helper T cells and activated memory CD4 + T cells in GVHD patients. Correlation analysis indicated positive associations of RUVBL1 and VAMP5 with resting NK cells, negative association of ABCA5 with resting NK cells, and negative correlations of GPR171, NMI, and SELPLG with resting mast cells.
CONCLUSIONS: This study developed a stable and interpretable GVHD prediction model, identified a set of potential immune-related biomarkers, and provided new insights into the immune mechanisms underlying GVHD.
METHODS: This study combined transcriptomic profiles from several GEO datasets of leukemia patients after transplantation. Feature genes were selected through expression pattern analysis combined with unsupervised clustering. Machine learning algorithms were applied to construct predictive models for GVHD, with model performance evaluated by ROC curves in both training and external validation cohorts. Additionally, functional enrichment analyses, including GO, KEGG, GSEA, and GSVA, were performed to investigate potential biological mechanisms. In the final step, we explored the patterns of immune cell infiltration and their relationships with the core genes.
RESULTS: A machine learning model based on 17 core genes was developed, with Ridge regression showing the best performance (AUC = 0.955). Key genes included HLA-DQA1, IL2RA, SELPLG, GPR65, and RUVBL1. Functional enrichment revealed involvement in immune regulation, cytokine production, and hematopoietic pathways. GSEA and GSVA analyses demonstrated significant activation of pathways such as inositol phosphate metabolism, ABC transporters, and Th1/Th2 cell differentiation in GVHD patients. Analysis of immune infiltration revealed elevated levels of follicular helper T cells and activated memory CD4 + T cells in GVHD patients. Correlation analysis indicated positive associations of RUVBL1 and VAMP5 with resting NK cells, negative association of ABCA5 with resting NK cells, and negative correlations of GPR171, NMI, and SELPLG with resting mast cells.
CONCLUSIONS: This study developed a stable and interpretable GVHD prediction model, identified a set of potential immune-related biomarkers, and provided new insights into the immune mechanisms underlying GVHD.